Merge pull request #4925 from Brisbanehuang/feat/group-profit-control

feat(scheduler): 分组级利润控制——按账号倍率过滤 token 调度候选

.github/workflows/backend-ci.yml

@@ -16,6 +16,9 @@ jobs:
         run: |
           /bin/bash -n deploy/apple-container.sh
           /bin/bash deploy/tests/apple-container-test.sh
+          /bin/sh deploy/tests/docker-compose-security-test.sh
+          /bin/sh deploy/tests/docker-runtime-resources-test.sh
+          /bin/sh deploy/test-caddyfile-cache.sh
 
   test:
     runs-on: ubuntu-latest

.goreleaser.simple.yaml

@@ -49,6 +49,7 @@ dockers:
     use: buildx
     extra_files:
       - deploy/docker-entrypoint.sh
+      - backend/resources
     build_flag_templates:
       - "--platform=linux/amd64"
       - "--label=org.opencontainers.image.version={{ .Version }}"

.goreleaser.yaml

@@ -65,6 +65,7 @@ dockers:
     use: buildx
     extra_files:
       - deploy/docker-entrypoint.sh
+      - backend/resources
     build_flag_templates:
       - "--platform=linux/amd64"
       - "--label=org.opencontainers.image.version={{ .Version }}"
@@ -80,6 +81,7 @@ dockers:
     use: buildx
     extra_files:
       - deploy/docker-entrypoint.sh
+      - backend/resources
     build_flag_templates:
       - "--platform=linux/arm64"
       - "--label=org.opencontainers.image.version={{ .Version }}"
@@ -95,6 +97,7 @@ dockers:
     use: buildx
     extra_files:
       - deploy/docker-entrypoint.sh
+      - backend/resources
     build_flag_templates:
       - "--platform=linux/amd64"
       - "--label=org.opencontainers.image.version={{ .Version }}"
@@ -110,6 +113,7 @@ dockers:
     use: buildx
     extra_files:
       - deploy/docker-entrypoint.sh
+      - backend/resources
     build_flag_templates:
       - "--platform=linux/arm64"
       - "--label=org.opencontainers.image.version={{ .Version }}"

Dockerfile.goreleaser

@@ -42,8 +42,9 @@ RUN addgroup -g 1000 sub2api && \
 
 WORKDIR /app
 
-# Copy pre-built binary from GoReleaser
+# Copy the pre-built binary and runtime fallback resources from GoReleaser
 COPY sub2api /app/sub2api
+COPY --chown=sub2api:sub2api backend/resources /app/resources
 
 # Create data directory
 RUN mkdir -p /app/data && chown -R sub2api:sub2api /app

README.md

@@ -59,8 +59,8 @@ Please read the following carefully before using this project:
 </tr>
 
 <tr>
-<td width="180"><a href="https://www.aicodemirror.com/register?invitecode=KMVZQM"><img src="assets/partners/logos/AICodeMirror.jpg" alt="AICodeMirror" width="150"></a></td>
-<td>Thanks to AICodeMirror for sponsoring this project! AICodeMirror provides official high-stability relay services for Claude Code / Codex / Gemini CLI, with enterprise-grade concurrency, fast invoicing, and 24/7 dedicated technical support. Claude Code / Codex / Gemini official channels at 38% / 2% / 9% of original price, with extra discounts on top-ups! AICodeMirror offers special benefits for sub2api users: register via <a href="https://www.aicodemirror.com/register?invitecode=KMVZQM">this link</a> to enjoy 20% off your first top-up, and enterprise customers can get up to 25% off!</td>
+<td width="180"><a href="https://www.aicodemirror.ai/register?invitecode=KMVZQM"><img src="assets/partners/logos/AICodeMirror.jpg" alt="AICodeMirror" width="150"></a></td>
+<td>Thanks to AICodeMirror for sponsoring this project! AICodeMirror provides official high-stability relay services for Claude Code / Codex / Gemini CLI, with enterprise-grade concurrency, fast invoicing, and 24/7 dedicated technical support. Claude Code / Codex / Gemini official channels at 38% / 2% / 9% of original price, with extra discounts on top-ups! AICodeMirror offers special benefits for sub2api users: register via <a href="https://www.aicodemirror.ai/register?invitecode=KMVZQM">this link</a> to enjoy 20% off your first top-up, and enterprise customers can get up to 25% off!</td>
 </tr>
 
 <tr>
@@ -84,12 +84,6 @@ Please read the following carefully before using this project:
 </td>
 </tr>
 
-<tr>
-<td width="180"><a href="https://unity2.ai/register?source=sub2api"><img src="assets/partners/logos/unity2.png" alt="unity2" width="150"></a></td>
-<td>Thanks to Unity2 for sponsoring this project! <a href="https://unity2.ai/register?source=sub2api">Unity2</a> is a high-performance AI model API relay for individuals, teams, and enterprises, handling 30B+ tokens/day with 5000 RPM concurrency. One API Key works across Claude Code, Codex, OpenAI models, IDE plugins, and Agent workflows, with balance billing, bundled subscriptions, enterprise invoicing, and 1-on-1 support. <a href="https://unity2.ai/register?source=sub2api">Register</a> to claim $2 in balance, plus $10 more by joining the official group — up to $12 in free credit.
-</td>
-</tr>
-
 <tr>
 <td width="180"><a href="https://veilx.io/#/hello/SJRBRVDV"><img src="assets/partners/logos/veilx.png" alt="veilx" width="150"></a></td>
 <td>Thanks to Veilx for sponsoring this project! <a href="https://veilx.io/#/hello/SJRBRVDV">Veilx</a> CDN is purpose-built for large-scale AI API traffic, deeply optimized for relay services and call chains across OpenAI, Claude, Gemini, and scenarios like chat, image generation, embeddings, and streaming — delivering lower latency and higher stability under heavy concurrency. It also offers China three-network optimized return lines, making it ideal for global AI relay platforms, overseas AI SaaS, and cross-border high-concurrency deployments.
@@ -102,45 +96,12 @@ Please read the following carefully before using this project:
 </td>
 </tr>
 
-<tr>
-<td width="180"><a href="https://apikl.ai"><img src="assets/partners/logos/apikl.png" alt="apikl" width="150"></a></td>
-<td>Thanks to Apikl for sponsoring this project! Built on Sub2API, the platform provides developers with relay services for Codex / Claude series models, focusing on long-term stability, high-speed direct connections, and excellent cost-effectiveness. It offers pay-as-you-go balance billing, enterprise-grade official invoices, and one-on-one dedicated support. <a href="https://apikl.ai">Register now</a> for a 1:1 top-up bonus — double your balance!
-</td>
-</tr>
-
-<tr>
-<td width="180"><a href="https://tokeneum.ai"><img src="assets/partners/logos/tokeneum.png" alt="tokeneum" width="150"></a></td>
-<td>Thanks to TokenEum for sponsoring this project! <a href="https://tokeneum.ai">TokenEum</a> is a comprehensive AI model aggregation platform and intelligent agent development company. It brings together top-tier international models — including Claude, Gemini, and OpenAI — alongside leading open-source models such as GLM, Qwen, and Kimi, offering a wide range of options across different quality and price tiers to suit every need. TokenEum also provides access to cutting-edge video generation models like Seedance2.0 and Happy Horse. Committed to transparency and honest business practices, TokenEum ensures all model information is accurate and reliable. Visit <a href="https://tokeneum.ai">tokeneum.ai</a> to get started.
-</td>
-</tr>
-
-<tr>
-<td width="180"><a href="https://666api.work/sub2api"><img src="assets/partners/logos/666api.jpg" alt="666api" width="150"></a></td>
-<td>Thanks to 666api for sponsoring this project! <a href="https://666api.work/sub2api">666api</a> is an all-in-one platform offering:<br>
-⚡ API Relay — Pay-as-you-go access to global models sourced 100% from official providers, up to 75% off official pricing<br>
-&nbsp;&nbsp;&nbsp;&nbsp;Exclusive: Zhipu GLM 50% off · DeepSeek V4-pro 50% off · Seedance 2.0 8% off (whitelisted) · HappyHorse Overseas 30% off (whitelisted)<br>
-🔑 GPT Subscription Accounts (same-origin IP included) · Global Residential IP <br>
-💰 Invoices supported
-</td>
-</tr>
-
-<tr>
-<td width="180"><a href="https://dis.chatdesks.cn/chatdesk/hsyqsub2api.html"><img src="assets/partners/logos/byteplus.png" alt="BytePlus" width="150"></a></td>
-<td>Thanks to Dola seed for sponsoring this project! Dola Seed 2.0 is a full‑modal general large model independently developed by ByteDance for the global market. Built on a unified multimodal architecture, it supports joint understanding and generation of text, images, audio, and video. It natively enables agent collaboration, with strong reasoning, long‑task execution, tool integration, and coding capabilities. It is widely applicable to smart cockpits, personal assistants, education, customer support, marketing, retail, and other scenarios. It excels in multimodal perception, end‑to‑end complex task delivery, stable interaction, and data security, and is readily accessible and deployable via the ModelArk platform.Register via <a href="https://dis.chatdesks.cn/chatdesk/hsyqsub2api.html">this link</a> to get 500,000 tokens of free inference quota per model.<a href="https://dis.chatdesks.cn/chatdesk/hsyqsub2api.html"> >>中国大陆地区的开发者请点击这里</a></td>
-</tr>
-
 <tr>
 <td width="180"><a href="https://sui-xiang.com/"><img src="assets/partners/logos/sui-xiang.jpg" alt="sui-xiang" width="150"></a></td>
 <td>Thanks to Suixiang AI Gateway for sponsoring this project! <a href="https://sui-xiang.com/">Suixiang AI Gateway</a> is a reliable and efficient API relay service provider offering relay services for Claude, Codex, Gemini, and more. A privacy-focused relay — no data reselling, no model dilution; privacy, transparency, and lightning-fast after-sales support. New accounts get ¥0.5 in trial credit daily by signing in; top-ups are 1:1, no subscription required, pay-as-you-go. Multi-line redundancy, cross-region disaster recovery, automatic failover, and uninterrupted long-link SSE. 99.9% availability — critical calls never fall behind.
 </td>
 </tr>
 
-<tr>
-<td width="180"><a href="https://www.miyaip.com/?invitecode=sub2api"><img src="assets/partners/logos/miyaip.png" alt="miyaip" width="150"></a></td>
-<td>Thanks to MiyaIP for sponsoring this project! <a href="https://www.miyaip.com/?invitecode=sub2api">MiyaIP</a> is a platform dedicated to global residential proxy network services, committed to providing high-quality, pure overseas residential IP resources for enterprise developers, cross-border business teams, and AI application users. It delivers stable, independent overseas network environments for AI platforms, overseas SaaS, and other online services, supporting multi-region access testing and project environment isolation. Ideal for development and testing scenarios that require access to overseas AI services, such as: AI model platform access, AI development testing, AI SaaS service usage, AI API debugging, and multi-region network environment validation.
-</td>
-</tr>
-
 <tr>
 <td width="180"><a href="https://anpin.ai"><img src="assets/partners/logos/anpin.jpg" alt="anpin" width="150"></a></td>
 <td>Thanks to <a href="https://anpin.ai">anpin.ai</a> for sponsoring this project! anpin.ai is a premium AI relay service platform dedicated to advancing AI accessibility. With an advanced technical architecture and globally distributed deployment, it provides users with a direct high-speed channel to the world's top-tier large language models.<br>
@@ -172,8 +133,8 @@ Model authenticity: no content intervention or secondary filtering — experienc
 </tr>
 
 <tr>
-<td width="180"><a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z?utm_source=zcf&utm_medium=partner&utm_campaign=zcf_2026&utm_content=default"><img src="assets/partners/logos/claudeapi.jpg" alt="claudeapi" width="150"></a></td>
-<td>Thanks to Claude API for sponsoring this project! <a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z?utm_source=zcf&utm_medium=partner&utm_campaign=zcf_2026&utm_content=default">Claude API</a> is an official-channel API provider focused on Claude models. Built on official Anthropic keys and the official AWS Bedrock channel, it delivers a stable integration experience for Claude Code and Agent applications, supports the full Claude model lineup, and retains official capabilities such as Tool Use and long context. The service involves no reverse engineering and no model degradation, making it a great fit for heavy Claude Code users, Agent engineers, and enterprise engineering teams. Register via the <a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z?utm_source=zcf&utm_medium=partner&utm_campaign=zcf_2026&utm_content=default">exclusive link</a> and contact customer support to receive free trial credits; invoicing and team onboarding are also supported.
+<td width="180"><a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z"><img src="assets/partners/logos/claudeapi.jpg" alt="claudeapi" width="150"></a></td>
+<td>Thanks to Claude API for sponsoring this project! <a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z">Claude API</a> is an official-channel API provider focused on Claude models. Built on official Anthropic keys and the official AWS Bedrock channel, it delivers a stable integration experience for Claude Code and Agent applications, supports the full Claude model lineup, and retains official capabilities such as Tool Use and long context. The service involves no reverse engineering and no model degradation, making it a great fit for heavy Claude Code users, Agent engineers, and enterprise engineering teams. Register via the <a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z">exclusive link</a> and contact customer support to receive free trial credits; invoicing and team onboarding are also supported.
 </td>
 </tr>
 
@@ -188,6 +149,26 @@ Model authenticity: no content intervention or secondary filtering — experienc
 <td><a href="https://nagora.ai/">Nagora</a> is a multi-model AI API gateway built for developers and teams. With a single account and API key, you can access more than 26 leading text and image models through one unified interface. It is compatible with OpenAI, Anthropic, and Gemini protocols and integrates seamlessly with development tools such as Claude Code, Codex, and Gemini CLI. The platform provides intelligent routing, automatic failover, transparent pricing, and consolidated billing, along with budget management, rate limiting, and concurrency controls. This makes AI usage more reliable and manageable across individual development, team collaboration, and production environments. No changes to your existing application are required. Simply replace the Base URL and API key to complete the integration in as little as one minute.</td>
 </tr>
 
+<tr>
+<td width="180"><a href="https://www.novada.com/?sub2api/"><img src="assets/partners/logos/novada.png" alt="Novada" width="150"></a></td>
+<td>Thanks to <a href="https://www.novada.com/?sub2api/">Novada</a> for sponsoring this project! Novada provides residential, ISP, datacenter, and mobile proxies, along with Web Unlocker and Scraper APIs for developers building AI applications and automation workflows. With global IP coverage, flexible rotating and sticky sessions, and precise geo-targeting, Novada helps teams access web data reliably for AI agent workflows, cross-region testing, web research, and browser automation. Explore Novada to build more stable and scalable AI workflows.</td>
+</tr>
+
+<tr>
+<td width="180"><a href="https://s.qiniu.com/u6rQrq"><img src="assets/partners/logos/qiniu.jpg" alt="Qiniu AI" width="150"></a></td>
+<td>Thanks to Qiniu AI for sponsoring this project! Qiniu AI is the enterprise-grade large-model MaaS platform under Qiniu Cloud (02567.HK), offering one-stop access to 150+ mainstream models worldwide, compatible with the protocols of major global model providers, and covering full-modality capabilities including text, image, audio, video, and file processing, serving over 1.69 million enterprises and developers. Qiniu AI offers an exclusive benefit for Sub2API users: register via <a href="https://s.qiniu.com/u6rQrq">this link</a> — enterprise users get 12 million tokens free, and developers get 3 million tokens free.</td>
+</tr>
+
+<tr>
+<td width="180"><a href="https://api.fenno.ai/s/dC4k"><img src="assets/partners/logos/fennoai.jpg" alt="FennoAI" width="150"></a></td>
+<td>Thanks to FennoAI for sponsoring this project! FennoAI is a high-stability, high-performance API relay provider for enterprise R&D teams and developers, compatible with the OpenAI and Anthropic protocols and seamlessly integrating with mainstream AI coding tools such as Codex, Claude Code, and OpenCode. The platform delivers enterprise-grade stability, supporting call volumes of 100 billion tokens per day, and supports business-to-business settlement and invoicing for both domestic and overseas entities to meet enterprise R&D and procurement needs. As an exclusive benefit for Sub2API users, purchase a subscription via the <a href="https://api.fenno.ai/s/dC4k">exclusive link</a> to get $50 worth of Coding Plan credit for only $1.99. Referral rewards are also available: invite friends to purchase and earn up to 20% commission — the more you invite, the more you earn.</td>
+</tr>
+
+<tr>
+<td width="180"><a href="https://lanox.ai/?c=6"><img src="assets/partners/logos/lanox.jpg" alt="LanoX AI" width="150"></a></td>
+<td>Thank you to LanoX AI for sponsoring this project! <a href="https://lanox.ai/?c=6">LanoX AI</a> provides stable, cost-effective global model access services for developers, teams, and enterprises. 🎁 New User Benefits — Claim millions of free tokens, plus 500+ free models for easier low-cost testing, validation, and deployment 🧠 Global Leading Models — GPT · Claude · Gemini · Qwen · Grok... 🎬 Multimodal Creation — Seedance 2.0 · GPT Image · Gemini Nano Banana 🛡️ Enterprise-Grade Reliability — High availability 💎 native capability output 💎 no intelligence degradation 💎 no model mixing 💎 transparent usage and billing 💎 💰 Lower API Costs — Top-tier models from as low as 10% of official pricing, with clear documentation, simple integration, invoicing support, and enterprise-scale batch usage 🏢 Enterprise Choice — Ideal for AI products, Agents, content platforms, and R&D teams with high-volume model usage</td>
+</tr>
+
 </table>
 
 ## Overview

README_CN.md

@@ -60,8 +60,8 @@
 </tr>
 
 <tr>
-<td width="180"><a href="https://www.aicodemirror.com/register?invitecode=KMVZQM"><img src="assets/partners/logos/AICodeMirror.jpg" alt="AICodeMirror" width="150"></a></td>
-<td>感谢 AICodeMirror 赞助了本项目！AICodeMirror 提供 Claude Code / Codex / Gemini CLI 官方高稳定性中转服务，企业级并发、快速开票、7×24 小时专属技术支持。Claude Code / Codex / Gemini 官方通道低至原价 38% / 2% / 9%，充值更享额外折扣！AICodeMirror 为 sub2api 用户提供专属福利：通过<a href="https://www.aicodemirror.com/register?invitecode=KMVZQM">此链接</a>注册，首次充值立享 8 折优惠，企业客户最高可享 75 折！</td>
+<td width="180"><a href="https://www.aicodemirror.ai/register?invitecode=KMVZQM"><img src="assets/partners/logos/AICodeMirror.jpg" alt="AICodeMirror" width="150"></a></td>
+<td>感谢 AICodeMirror 赞助了本项目！AICodeMirror 提供 Claude Code / Codex / Gemini CLI 官方高稳定性中转服务，企业级并发、快速开票、7×24 小时专属技术支持。Claude Code / Codex / Gemini 官方通道低至原价 38% / 2% / 9%，充值更享额外折扣！AICodeMirror 为 sub2api 用户提供专属福利：通过<a href="https://www.aicodemirror.ai/register?invitecode=KMVZQM">此链接</a>注册，首次充值立享 8 折优惠，企业客户最高可享 75 折！</td>
 </tr>
 
 <tr>
@@ -87,12 +87,6 @@
 </td>
 </tr>
 
-<tr>
-<td width="180"><a href="https://unity2.ai/register?source=sub2api"><img src="assets/partners/logos/unity2.png" alt="unity2" width="150"></a></td>
-<td>感谢 Unity2 赞助本项目！ <a href="https://unity2.ai/register?source=sub2api">Unity2</a> 是面向个人开发者、团队、企业的高性能 AI 模型 API 中转平台，长期服务国内头部企业，日均承载超 300 亿 token 调用，支持 5000 RPM 级高并发。一个 API Key 即可适配 Claude Code、Codex、OpenAI 模型、IDE 插件和 Agent 工作流等场景。具备企业级稳定供应能力，在高并发、持续调用和团队集中采购场景下依然保持低延迟、高可用。同时支持余额计费、组合订阅、首充优惠、企业开票、专属 1v1 对接，适合个人高频使用和企业长期接入。现在注册 Unity2.ai 可领取 $2 余额，加入官方群再送 $10 余额，合计最高可领 $12 免费额度，适合先体验后长期使用。<a href="https://unity2.ai/register?source=sub2api">注册链接</a>
-</td>
-</tr>
-
 <tr>
 <td width="180"><a href="https://veilx.io/#/hello/SJRBRVDV"><img src="assets/partners/logos/veilx.png" alt="veilx" width="150"></a></td>
 <td>感谢 Veilx 赞助本项目！ <a href="https://veilx.io/#/hello/SJRBRVDV">Veilx</a> CDN 专为超大规模 API 请求场景打造，针对 AI 中转站业务与 AI API 调用链路进行了深度优化，轻松应对高并发、高频请求与大流量传输，为开发者与企业提供更快、更稳、更低延迟的加速体验。无论是 OpenAI、Claude、Gemini 等 AI 接口中转，还是聊天、绘图、Embedding、流式输出等复杂场景，Veilx 都能显著提升响应速度与连接稳定性，有效降低网络波动带来的超时与失败问题。同时，Veilx 提供中国三网优化回国极速线路，大幅提升中国大陆地区访问海外 AI 服务的速度与稳定性，特别适合全球 AI 中转平台、海外 AI SaaS、跨境业务与高并发 API 系统部署。专为 AI API 而生，让你的 AI 中转服务更快、更稳、更省心。<a href="https://veilx.io/#/hello/SJRBRVDV">购买地址</a>
@@ -105,45 +99,12 @@
 </td>
 </tr>
 
-<tr>
-<td width="180"><a href="https://apikl.ai"><img src="assets/partners/logos/apikl.png" alt="apikl" width="150"></a></td>
-<td>感谢 Apikl 赞助本项目！平台基于 Sub2API 搭建，为开发者提供 Codex / Claude 系列模型的中转服务，专注于长期稳定、高速直连与高性价比。支持按量计费的余额结算、企业级正规发票及一对一专属对接。<a href="https://apikl.ai">立即注册</a>即享充值 1:1 赠送 — 余额翻倍！
-</td>
-</tr>
-
-<tr>
-<td width="180"><a href="https://tokeneum.ai"><img src="assets/partners/logos/tokeneum.png" alt="tokeneum" width="150"></a></td>
-<td>感谢 TokenEum 赞助本项目！<a href="https://tokeneum.ai">TokenEum</a> 是一家综合性 AI 模型聚合平台与智能体开发公司，汇聚 Claude、Gemini、OpenAI 等国际顶级模型，以及 GLM、Qwen、Kimi 等主流开源模型，提供不同质量与价格梯度的丰富选择，满足多样化需求。平台还接入了 Seedance2.0、Happy Horse 等前沿视频生成模型。秉持透明诚信的经营理念，TokenEum 确保所有模型信息真实可靠。访问 <a href="https://tokeneum.ai">tokeneum.ai</a> 开始使用。
-</td>
-</tr>
-
-<tr>
-<td width="180"><a href="https://666api.work/sub2api"><img src="assets/partners/logos/666api.jpg" alt="666api" width="150"></a></td>
-<td>感谢 666api 赞助本项目！<a href="https://666api.work/sub2api">666api</a> 是一站式综合服务平台，提供：<br>
-⚡ API 中转 — 全球模型按量计费接入，100% 官方源直供，最高 75 折优惠<br>
-&nbsp;&nbsp;&nbsp;&nbsp;独家特惠：智谱 GLM 5 折 · DeepSeek V4-pro 5 折 · Seedance 2.0 0.8 折（白名单）· HappyHorse 海外版 3 折（白名单）<br>
-🔑 GPT 订阅账号（含同源 IP）· 全球住宅 IP<br>
-💰 支持开票
-</td>
-</tr>
-
-<tr>
-<td width="180"><a href="https://dis.chatdesks.cn/chatdesk/hsyqsub2api.html"><img src="assets/partners/logos/huoshan.png" alt="HuoShan" width="150"></a></td>
-<td>感谢火山方舟 Agent Plan 模型赞助了本项目！方舟 Agent Plan 模型订阅套餐集成了包含 Doubao-Seed、Doubao-Seedance、Doubao-Seedream 等在内的字节跳动自研 SOTA 级模型，覆盖文本、代码、图像、视频等多模态任务。最新支持 MiniMax-M3、DeepSeek-V4 系列、GLM-5.1、Doubao-Seed-2.0 系列、Kimi-K2.6 等模型，工具不限。超全模态模型与 Harness 升级一步到位，深度支持 Agent 框架与 AI 编程工具。一次订阅，可以为不同任务切换合适的 AI 引擎。方舟 Coding Plan 为 Sub2Api 的用户提供了专属福利：通过<a href="https://dis.chatdesks.cn/chatdesk/hsyqsub2api.html">此链接</a>订阅方舟 Coding Plan，新客户首两个月享 2.5 折优惠 <a href="https://dis.chatdesks.cn/chatdesk/hsyqsub2api.html">>>For developers outside Mainland China, please click here</a></td>
-</tr>
-
 <tr>
 <td width="180"><a href="https://sui-xiang.com/"><img src="assets/partners/logos/sui-xiang.jpg" alt="sui-xiang" width="150"></a></td>
 <td>感谢 随想AI网关 赞助本项目！<a href="https://sui-xiang.com/">随想AI网关</a>  是一家可靠高效的 API 中继服务提供商，提供 Claude、Codex、Gemini 等的中继服务。注重隐私的中转站·无数据倒卖·无模型掺水，隐私，透明，极速售后。新账户注册每日签到就送 0.5 元测试额度，充值额度 1:1，无需订阅，按量付费。多线路冗余、跨区域容灾、自动故障切换,长链路 SSE 不中断。99.9% 可用性,关键调用从不掉队。
 </td>
 </tr>
 
-<tr>
-<td width="180"><a href="https://www.miyaip.com/?invitecode=sub2api"><img src="assets/partners/logos/miyaip.png" alt="miyaip" width="150"></a></td>
-<td>感谢 MiyaIP 赞助本项目！<a href="https://www.miyaip.com/?invitecode=sub2api">MiyaIP</a> 是一家专注于全球住宅代理网络服务的平台，致力于为企业开发者、跨境业务团队及AI 应用用户提供高质量、纯净的海外住宅 IP 资源。为 AI 平台、海外 SaaS 及其他在线服务提供稳定、独立的海外网络环境，支持多地区访问测试和项目环境隔离。适用于需要访问海外 AI 服务的开发和测试场景，例如：AI 模型平台访问、AI 开发测试、AI SaaS 服务使用、AI API 调试、多地区网络环境验证
-</td>
-</tr>
-
 <tr>
 <td width="180"><a href="https://anpin.ai"><img src="assets/partners/logos/anpin.jpg" alt="anpin" width="150"></a></td>
 <td>感谢 <a href="https://anpin.ai">anpin.ai</a> 赞助本项目！anpin.ai 是一家致力于推动 AI 普惠的高端 AI 中转服务平台。我们以先进的技术架构和全球分布式部署，为用户提供直达国际顶尖大模型的高速通道。<br>
@@ -175,8 +136,8 @@
 </tr>
 
 <tr>
-<td width="180"><a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z?utm_source=zcf&utm_medium=partner&utm_campaign=zcf_2026&utm_content=default"><img src="assets/partners/logos/claudeapi.jpg" alt="claudeapi" width="150"></a></td>
-<td>感谢 Claude API 对本项目的赞助！ <a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z?utm_source=zcf&utm_medium=partner&utm_campaign=zcf_2026&utm_content=default">Claude API</a> 是专注 Claude 模型的官方渠道 API 服务商，基于 Anthropic 官方 Key 与 AWS Bedrock 官方渠道，提供稳定的 Claude Code 与 Agent 应用接入体验，支持 Claude 全系列模型，保留 Tool Use、长上下文等官方能力。服务非逆向、非降智，适合 Claude Code 深度用户、Agent 工程师与企业技术团队使用。通过<a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z?utm_source=zcf&utm_medium=partner&utm_campaign=zcf_2026&utm_content=default">[专属链接]</a>注册后联系客服，可领取免费测试额度，并支持开票和团队对接。
+<td width="180"><a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z"><img src="assets/partners/logos/claudeapi.jpg" alt="claudeapi" width="150"></a></td>
+<td>感谢 Claude API 对本项目的赞助！ <a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z">Claude API</a> 是专注 Claude 模型的官方渠道 API 服务商，基于 Anthropic 官方 Key 与 AWS Bedrock 官方渠道，提供稳定的 Claude Code 与 Agent 应用接入体验，支持 Claude 全系列模型，保留 Tool Use、长上下文等官方能力。服务非逆向、非降智，适合 Claude Code 深度用户、Agent 工程师与企业技术团队使用。通过<a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z">[专属链接]</a>注册后联系客服，可领取免费测试额度，并支持开票和团队对接。
 </td>
 </tr>
 
@@ -191,6 +152,26 @@
 <td><a href="https://nagora.ai/">Nagora</a> 是专为开发者和团队打造的多模型 AI API 网关。通过一个账户和一枚 API Key，即可统一调用 26+ 款主流文本与图像模型，兼容 OpenAI、Anthropic 与 Gemini 协议，并可无缝接入 Claude Code、Codex、Gemini CLI 等开发工具。平台提供智能路由、自动故障转移、透明计费与统一账单，同时支持预算、限速、并发控制，让个人开发、团队协作和生产环境中的 AI 调用更稳定、更可控。无需改造现有应用，只需替换 Base URL 与 API Key，最快 1 分钟即可完成接入。</td>
 </tr>
 
+<tr>
+<td width="180"><a href="https://www.novada.com/?sub2api/"><img src="assets/partners/logos/novada.png" alt="Novada" width="150"></a></td>
+<td>感谢 <a href="https://www.novada.com/?sub2api/">Novada</a> 赞助本项目！Novada 为构建 AI 应用与自动化工作流的开发者提供住宅代理、ISP 代理、数据中心代理与移动代理，以及 Web Unlocker 和 Scraper API。凭借全球 IP 覆盖、灵活的轮换与粘性会话以及精准的地理定位，Novada 帮助团队在 AI Agent 工作流、跨区域测试、网络调研与浏览器自动化等场景中稳定获取网络数据。立即体验 Novada，构建更稳定、更可扩展的 AI 工作流。</td>
+</tr>
+
+<tr>
+<td width="180"><a href="https://s.qiniu.com/u6rQrq"><img src="assets/partners/logos/qiniu.jpg" alt="七牛云AI" width="150"></a></td>
+<td>感谢 七牛云AI 赞助本项目！七牛云AI 是七牛云（02567.HK）旗下企业级大模型 MaaS 平台，一站式调用全球 150+ 主流模型，兼容全球主流模型厂商协议，覆盖文本、图像、音频、视频、文件处理等全模态处理能力，服务超过169万企业及开发者用户。七牛云 AI 为 Sub2API 的用户提供了专属福利：通过<a href="https://s.qiniu.com/u6rQrq">此链接</a>注册，企业用户免费领1200万Token，开发者免费领300万Token。</td>
+</tr>
+
+<tr>
+<td width="180"><a href="https://api.fenno.ai/s/dC4k"><img src="assets/partners/logos/fennoai.jpg" alt="FennoAI" width="150"></a></td>
+<td>感谢 FennoAI 赞助本项目！FennoAI 是一家面向企业研发团队和开发者的高稳定、高性能 API 中转服务商，兼容 OpenAI 与 Anthropic 协议，可无缝接入 Codex、Claude Code、OpenCode 等主流 AI 编程工具。平台具备企业级稳定性，可支撑千亿 Token/日的调用规模，并支持境内外主体公对公结算及开票，满足企业级研发与采购需求。作为 Sub2API 用户专属福利，通过<a href="https://api.fenno.ai/s/dC4k">专属链接</a>购买订阅，仅需 1.99 美元即可获得价值 50 美元的 Coding Plan 额度。同时支持邀请奖励，邀请好友购买最高可获得 20% 返佣，邀请越多，奖励越高。</td>
+</tr>
+
+<tr>
+<td width="180"><a href="https://lanox.ai/?c=6"><img src="assets/partners/logos/lanox.jpg" alt="LanoX AI" width="150"></a></td>
+<td>感谢 LanoX AI 对本项目的赞助！<a href="https://lanox.ai/?c=6">LanoX AI</a> 为开发者、团队与企业提供稳定、高性价比的全球模型接入服务。 🎁 新用户福利 — 免费领取 百万 Token ,更有500+ 免费模型 — 低成本测试、验证、部署更轻松 🧠 全球主流模型 — GPT · Claude · Gemini · Qwen · Grok... 🎬 多模态创作 — Seedance 2.0 · GPT Image · Gemini Nano Banana 🛡️ 企业级稳定服务 — 高可用💎原生能力输出💎不降智💎不混模💎调用与计费透明💎 💰 更低调用成本 — 顶级模型低至官方价 1 折起，文档清晰、接入简单、支持开票与企业批量调用 🏢 企业优选 — 适用于 AI 产品、Agent、内容平台、研发团队批量调用</td>
+</tr>
+
 </table>
 
 ## 项目概述

README_JA.md

@@ -59,8 +59,8 @@
 </tr>
 
 <tr>
-<td width="180"><a href="https://www.aicodemirror.com/register?invitecode=KMVZQM"><img src="assets/partners/logos/AICodeMirror.jpg" alt="AICodeMirror" width="150"></a></td>
-<td>AICodeMirror のご支援に感謝します！AICodeMirror は Claude Code / Codex / Gemini CLI の公式高安定性リレーサービスを提供しており、エンタープライズグレードの同時実行、迅速な請求書発行、24時間年中無休の専属テクニカルサポートを備えています。Claude Code / Codex / Gemini の公式チャネルを定価の 38% / 2% / 9% で利用可能、チャージ時にはさらに追加割引！AICodeMirror は sub2api ユーザー向けに特別特典を提供中：<a href="https://www.aicodemirror.com/register?invitecode=KMVZQM">こちらのリンク</a>から登録すると、初回チャージが 20% オフ、法人のお客様は最大 25% オフ！</td>
+<td width="180"><a href="https://www.aicodemirror.ai/register?invitecode=KMVZQM"><img src="assets/partners/logos/AICodeMirror.jpg" alt="AICodeMirror" width="150"></a></td>
+<td>AICodeMirror のご支援に感謝します！AICodeMirror は Claude Code / Codex / Gemini CLI の公式高安定性リレーサービスを提供しており、エンタープライズグレードの同時実行、迅速な請求書発行、24時間年中無休の専属テクニカルサポートを備えています。Claude Code / Codex / Gemini の公式チャネルを定価の 38% / 2% / 9% で利用可能、チャージ時にはさらに追加割引！AICodeMirror は sub2api ユーザー向けに特別特典を提供中：<a href="https://www.aicodemirror.ai/register?invitecode=KMVZQM">こちらのリンク</a>から登録すると、初回チャージが 20% オフ、法人のお客様は最大 25% オフ！</td>
 </tr>
 
 <tr>
@@ -86,12 +86,6 @@
 </td>
 </tr>
 
-<tr>
-<td width="180"><a href="https://unity2.ai/register?source=sub2api"><img src="assets/partners/logos/unity2.png" alt="unity2" width="150"></a></td>
-<td>Unity2 のご支援に感謝します！<a href="https://unity2.ai/register?source=sub2api">Unity2</a> は個人開発者、チーム、企業向けの高性能 AI モデル API 中継プラットフォームです。中国の大手企業に長期にわたりサービスを提供しており、1日あたり 300 億以上のトークン呼び出しを処理し、5000 RPM 級の高並列性をサポートします。1つの API キーで Claude Code、Codex、OpenAI モデル、IDE プラグイン、Agent ワークフローなど様々なシナリオに対応できます。エンタープライズグレードの安定供給能力を備え、高並列・継続的な呼び出し・チームの集中購入シーンでも低レイテンシと高可用性を維持します。残高課金、組み合わせサブスクリプション、初回チャージ特典、企業向け請求書発行、専属 1v1 サポートにも対応しており、個人の頻繁な利用にも企業の長期導入にも適しています。今 Unity2.ai に登録すると $2 の残高、公式グループに参加するとさらに $10 の残高がもらえ、合計最大 $12 の無料クレジットを獲得できます — 試用後に長期利用したい方に最適です。<a href="https://unity2.ai/register?source=sub2api">登録リンク</a>
-</td>
-</tr>
-
 <tr>
 <td width="180"><a href="https://veilx.io/#/hello/SJRBRVDV"><img src="assets/partners/logos/veilx.png" alt="veilx" width="150"></a></td>
 <td>Veilx のご支援に感謝します！<a href="https://veilx.io/#/hello/SJRBRVDV">Veilx</a> CDN は超大規模 API リクエストシナリオ向けに設計されており、AI 中継サービスと AI API 呼び出しチェーンに対して深く最適化されています。高並列・高頻度リクエスト・大容量トラフィックに容易に対応し、開発者と企業により高速で安定した、低レイテンシの加速体験を提供します。OpenAI、Claude、Gemini などの AI インターフェース中継はもちろん、チャット、画像生成、Embedding、ストリーミング出力などの複雑なシナリオでも、Veilx は応答速度と接続安定性を大幅に向上させ、ネットワーク変動によるタイムアウトや失敗を効果的に削減します。さらに、Veilx は中国三大ネットワーク最適化の高速回線を提供しており、中国本土から海外 AI サービスへのアクセス速度と安定性を大幅に向上させます。グローバル AI 中継プラットフォーム、海外 AI SaaS、越境ビジネス、高並列 API システム展開に特に適しています。AI API のために生まれ、あなたの AI 中継サービスをより速く、より安定して、より安心に。<a href="https://veilx.io/#/hello/SJRBRVDV">購入リンク</a>
@@ -104,41 +98,12 @@
 </td>
 </tr>
 
-<tr>
-<td width="180"><a href="https://apikl.ai"><img src="assets/partners/logos/apikl.png" alt="apikl" width="150"></a></td>
-<td>Apikl のご支援に感謝します！Sub2API をベースに構築された本プラットフォームは、開発者向けに Codex / Claude シリーズモデルの中継サービスを提供しています。長期安定性、高速直結、高いコストパフォーマンスを重視し、従量課金の残高ベース課金、エンタープライズグレードの正規請求書、1対1の専属サポートを提供します。<a href="https://apikl.ai">今すぐ登録</a>でチャージ 1:1 ボーナス — 残高が倍に！
-</td>
-</tr>
-
-<tr>
-<td width="180"><a href="https://tokeneum.ai"><img src="assets/partners/logos/tokeneum.png" alt="tokeneum" width="150"></a></td>
-<td>TokenEum のご支援に感謝します！<a href="https://tokeneum.ai">TokenEum</a> は総合的な AI モデル集約プラットフォームおよびインテリジェントエージェント開発企業です。Claude、Gemini、OpenAI などの世界トップクラスのモデルに加え、GLM、Qwen、Kimi などの主要なオープンソースモデルも集約しており、品質と価格の異なる豊富な選択肢を提供してあらゆるニーズに対応します。また、Seedance2.0 や Happy Horse などの最先端の動画生成モデルも利用可能です。TokenEum は透明性と誠実なビジネスを重視し、すべてのモデル情報の正確性と信頼性を保証します。<a href="https://tokeneum.ai">tokeneum.ai</a> でぜひお試しください。
-</td>
-</tr>
-
-<tr>
-<td width="180"><a href="https://666api.work/sub2api"><img src="assets/partners/logos/666api.jpg" alt="666api" width="150"></a></td>
-<td>666api のご支援に感謝します！<a href="https://666api.work/sub2api">666api</a> はオールインワンプラットフォームで、以下を提供しています：⚡ API 中継 — グローバルモデルへの従量課金アクセス、100% 公式ソースから直接供給、公式価格の最大 75% オフ。独占特典：Zhipu GLM 50% オフ・DeepSeek V4-pro 50% オフ・Seedance2.0 8% オフ（ホワイトリスト）・HappyHorse Overseas 30% オフ（ホワイトリスト）🔑 GPT サブスクリプションアカウント — 同源 IP 込み・グローバル住宅 IP 💰 請求書発行対応
-</td>
-</tr>
-
-<tr>
-<td width="180"><a href="https://dis.chatdesks.cn/chatdesk/hsyqsub2api.html"><img src="assets/partners/logos/byteplus.png" alt="BytePlus" width="150"></a></td>
-<td>Dola seed のご支援に感謝します！Dola Seed 2.0 は、ByteDance がグローバル市場向けに独自開発した全モーダル汎用大規模モデルです。統一されたマルチモーダルアーキテクチャに基づき、テキスト・画像・音声・動画の共同理解と生成をサポートします。エージェント協調をネイティブに有効化し、強力な推論・長期タスク実行・ツール統合・コーディング能力を備えています。スマートコックピット、パーソナルアシスタント、教育、カスタマーサポート、マーケティング、リテールなど多様なシナリオに広く適用可能です。マルチモーダル知覚、エンドツーエンドの複雑タスク配信、安定したインタラクション、データセキュリティに優れ、ModelArk プラットフォームを通じて容易にアクセス・デプロイできます。<a href="https://dis.chatdesks.cn/chatdesk/hsyqsub2api.html">こちらのリンク</a>から登録すると、各モデルにつき 50 万トークンの無料推論クォータを獲得できます。<a href="https://dis.chatdesks.cn/chatdesk/hsyqsub2api.html"> >>中国大陸地域の開発者はこちらをクリック</a></td>
-</tr>
-
 <tr>
 <td width="180"><a href="https://sui-xiang.com/"><img src="assets/partners/logos/sui-xiang.jpg" alt="sui-xiang" width="150"></a></td>
 <td>随想AI ゲートウェイのご支援に感謝します！<a href="https://sui-xiang.com/">随想AI ゲートウェイ</a> は信頼性と効率に優れた API 中継サービスプロバイダーで、Claude、Codex、Gemini などの中継サービスを提供しています。プライバシー重視の中継ステーション・データ転売なし・モデル水増しなし、プライバシー・透明性・超高速アフターサービス。新規アカウント登録後、毎日サインインで 0.5 元のテストクレジットがもらえ、チャージは 1:1、サブスクリプション不要、従量課金。マルチライン冗長、クロスリージョン災害復旧、自動フェイルオーバー、長時間 SSE 接続が途切れません。99.9% の可用性、重要な呼び出しは決して遅れません。
 </td>
 </tr>
 
-<tr>
-<td width="180"><a href="https://www.miyaip.com/?invitecode=sub2api"><img src="assets/partners/logos/miyaip.png" alt="miyaip" width="150"></a></td>
-<td>MiyaIP のご支援に感謝します！<a href="https://www.miyaip.com/?invitecode=sub2api">MiyaIP</a> はグローバル住宅プロキシネットワークサービスに特化したプラットフォームで、企業開発者、越境ビジネスチーム、AI アプリケーションユーザーに高品質で純粋な海外住宅 IP リソースを提供することに注力しています。AI プラットフォーム、海外 SaaS、その他のオンラインサービスに安定した独立した海外ネットワーク環境を提供し、複数地域でのアクセステストやプロジェクト環境の分離をサポートします。海外 AI サービスへのアクセスが必要な開発・テストシナリオに最適です。例：AI モデルプラットフォームアクセス、AI 開発テスト、AI SaaS サービス利用、AI API デバッグ、複数地域のネットワーク環境検証など。
-</td>
-</tr>
-
 <tr>
 <td width="180"><a href="https://anpin.ai"><img src="assets/partners/logos/anpin.jpg" alt="anpin" width="150"></a></td>
 <td><a href="https://anpin.ai">anpin.ai</a> のご支援に感謝します！anpin.ai は AI の普及推進に取り組むハイエンド AI 中継サービスプラットフォームです。先進的な技術アーキテクチャとグローバル分散デプロイメントにより、国際トップクラスの大規模モデルへの高速直通ルートを提供します。<br>
@@ -170,8 +135,8 @@
 </tr>
 
 <tr>
-<td width="180"><a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z?utm_source=zcf&utm_medium=partner&utm_campaign=zcf_2026&utm_content=default"><img src="assets/partners/logos/claudeapi.jpg" alt="claudeapi" width="150"></a></td>
-<td>Claude API のご支援に感謝します！<a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z?utm_source=zcf&utm_medium=partner&utm_campaign=zcf_2026&utm_content=default">Claude API</a> は Claude モデルに特化した公式チャネルの API サービスプロバイダーで、Anthropic 公式キーと AWS Bedrock 公式チャネルをベースに、Claude Code や Agent アプリケーションへの安定した接続体験を提供します。Claude 全シリーズのモデルに対応し、Tool Use や長文コンテキストなどの公式機能もそのまま利用可能。リバースエンジニアリングやモデル劣化のないサービスで、Claude Code のヘビーユーザー、Agent エンジニア、企業の技術チームに最適です。<a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z?utm_source=zcf&utm_medium=partner&utm_campaign=zcf_2026&utm_content=default">専用リンク</a>から登録後カスタマーサポートへご連絡いただくと、無料お試しクレジットを受け取れます。請求書発行やチーム導入にも対応しています。
+<td width="180"><a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z"><img src="assets/partners/logos/claudeapi.jpg" alt="claudeapi" width="150"></a></td>
+<td>Claude API のご支援に感謝します！<a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z">Claude API</a> は Claude モデルに特化した公式チャネルの API サービスプロバイダーで、Anthropic 公式キーと AWS Bedrock 公式チャネルをベースに、Claude Code や Agent アプリケーションへの安定した接続体験を提供します。Claude 全シリーズのモデルに対応し、Tool Use や長文コンテキストなどの公式機能もそのまま利用可能。リバースエンジニアリングやモデル劣化のないサービスで、Claude Code のヘビーユーザー、Agent エンジニア、企業の技術チームに最適です。<a href="https://console.claudeapi.com/agent/register/drTKjyn6wGLK061Z">専用リンク</a>から登録後カスタマーサポートへご連絡いただくと、無料お試しクレジットを受け取れます。請求書発行やチーム導入にも対応しています。
 </td>
 </tr>
 
@@ -186,6 +151,26 @@
 <td><a href="https://nagora.ai/">Nagora</a>は、開発者やチーム向けに設計されたマルチモデルAI APIゲートウェイです。1つのアカウントと1つのAPIキーだけで、26種類以上の主要なテキストモデルおよび画像モデルを一元的に利用できます。OpenAI、Anthropic、Geminiの各プロトコルに対応し、Claude Code、Codex、Gemini CLIなどの開発ツールにもシームレスに接続できます。 プラットフォームには、インテリジェントルーティング、自動フェイルオーバー、透明性の高い料金体系、請求の一元管理に加え、予算管理、レート制限、同時実行数の制御機能が備わっています。これにより、個人開発、チームでの共同作業、本番環境におけるAI APIの利用を、より安定的かつ柔軟に管理できます。 既存のアプリケーションを改修する必要はありません。Base URLとAPIキーを置き換えるだけで、最短1分で導入を完了できます。</td>
 </tr>
 
+<tr>
+<td width="180"><a href="https://www.novada.com/?sub2api/"><img src="assets/partners/logos/novada.png" alt="Novada" width="150"></a></td>
+<td>Novada のご支援に感謝します！<a href="https://www.novada.com/?sub2api/">Novada</a> は、AI アプリケーションや自動化ワークフローを構築する開発者向けに、レジデンシャル、ISP、データセンター、モバイルプロキシに加え、Web Unlocker と Scraper API を提供しています。グローバルな IP カバレッジ、柔軟なローテーション／スティッキーセッション、精密なジオターゲティングにより、AI エージェントワークフロー、クロスリージョンテスト、Web リサーチ、ブラウザ自動化などのシーンで、チームが Web データへ確実にアクセスできるよう支援します。Novada で、より安定しスケーラブルな AI ワークフローを構築しましょう。</td>
+</tr>
+
+<tr>
+<td width="180"><a href="https://s.qiniu.com/u6rQrq"><img src="assets/partners/logos/qiniu.jpg" alt="Qiniu AI" width="150"></a></td>
+<td>七牛云AI のご支援に感謝します！七牛云AI は、七牛云（02567.HK）傘下のエンタープライズ向け大規模モデル MaaS プラットフォームです。世界の主要モデル 150+ をワンストップで利用でき、グローバル主要モデルプロバイダーのプロトコルに対応し、テキスト、画像、音声、動画、ファイル処理などのフルモーダル処理能力をカバー。169万を超える企業・開発者ユーザーにサービスを提供しています。Sub2API ユーザー向けの限定特典として、<a href="https://s.qiniu.com/u6rQrq">こちらのリンク</a>から登録すると、企業ユーザーは 1200万 Token、開発者は 300万 Token を無料で獲得できます。</td>
+</tr>
+
+<tr>
+<td width="180"><a href="https://api.fenno.ai/s/dC4k"><img src="assets/partners/logos/fennoai.jpg" alt="FennoAI" width="150"></a></td>
+<td>FennoAI のご支援に感謝します！FennoAI は、企業の研究開発チームと開発者向けの高安定・高性能 API 中継サービスプロバイダーです。OpenAI と Anthropic のプロトコルに対応し、Codex、Claude Code、OpenCode などの主要 AI コーディングツールにシームレスに接続できます。エンタープライズ級の安定性を備え、1 日あたり千億規模の Token 呼び出しに対応。国内外法人間の企業間決済と請求書発行もサポートし、企業の研究開発・調達ニーズに応えます。Sub2API ユーザー限定特典として、<a href="https://api.fenno.ai/s/dC4k">専用リンク</a>からサブスクリプションを購入すると、わずか 1.99 ドルで 50 ドル相当の Coding Plan クレジットを獲得できます。さらに招待報酬にも対応しており、友達の購入で最大 20% の還元を獲得可能。招待が多いほど、報酬も増えます。</td>
+</tr>
+
+<tr>
+<td width="180"><a href="https://lanox.ai/?c=6"><img src="assets/partners/logos/lanox.jpg" alt="LanoX AI" width="150"></a></td>
+<td>本プロジェクトをご支援いただいている <a href="https://lanox.ai/?c=6">LanoX AI</a> に感謝します！LanoX AI は、開発者、チーム、企業向けに、安定性とコストパフォーマンスに優れたグローバルモデル接続サービスを提供しています。 🎁 新規ユーザー特典 — 数百万 Token を無料で獲得可能。さらに 500+ の無料モデルで、低コストなテスト、検証、デプロイをより簡単に 🧠 世界の主要モデル — GPT · Claude · Gemini · Qwen · Grok... 🎬 マルチモーダル制作 — Seedance 2.0 · GPT Image · Gemini Nano Banana 🛡️ エンタープライズ級の安定性 — 高可用性 💎 ネイティブ能力の出力 💎 性能劣化なし 💎 モデル混在なし 💎 利用量と課金が透明 💎 💰 より低い利用コスト — トップモデルを公式価格の 1 割から利用可能。明確なドキュメント、簡単な接続、請求書発行、企業向け一括利用に対応 🏢 企業に最適 — AI プロダクト、Agent、コンテンツプラットフォーム、大量利用する開発チームに最適</td>
+</tr>
+
 </table>
 
 ## 概要

backend/cmd/profit-preview/main.go

@@ -0,0 +1,237 @@
+// profit-preview 读取生产只读导出的 JSON（分组利润配置、账号倍率与探测状态、
+// 用户覆盖倍率、主力模型清单），复用线上 U/D/阈值判定做五平台离线预演。
+//
+// 用法：
+//
+//	go run ./cmd/profit-preview -input dump.json [-assume-enabled] [-json]
+package main
+
+import (
+	"encoding/json"
+	"flag"
+	"fmt"
+	"os"
+	"sort"
+	"strings"
+	"time"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+)
+
+type inputGroup struct {
+	ID                   int64   `json:"id"`
+	Name                 string  `json:"name"`
+	Platform             string  `json:"platform"`
+	RateMultiplier       float64 `json:"rate_multiplier"`
+	SubscriptionType     string  `json:"subscription_type"`
+	ProfitControlEnabled bool    `json:"profit_control_enabled"`
+	ProfitMinMargin      float64 `json:"profit_min_margin"`
+	ProfitSafetyBuffer   float64 `json:"profit_safety_buffer"`
+	PeakRateEnabled      bool    `json:"peak_rate_enabled"`
+	PeakStart            string  `json:"peak_start"`
+	PeakEnd              string  `json:"peak_end"`
+	PeakRateMultiplier   float64 `json:"peak_rate_multiplier"`
+}
+
+type inputAccount struct {
+	ID             int64             `json:"id"`
+	Name           string            `json:"name"`
+	Platform       string            `json:"platform"`
+	Type           string            `json:"type"`
+	RateMultiplier *float64          `json:"rate_multiplier"`
+	Extra          map[string]any    `json:"extra"`
+	ModelMapping   map[string]string `json:"model_mapping"`
+}
+
+type inputEntry struct {
+	Group         inputGroup          `json:"group"`
+	Accounts      []inputAccount      `json:"accounts"`
+	UserOverrides map[string]*float64 `json:"user_overrides"`
+	Models        []string            `json:"models"`
+}
+
+type inputDoc struct {
+	Groups []inputEntry `json:"groups"`
+}
+
+func main() {
+	inputPath := flag.String("input", "", "生产只读导出 JSON 路径")
+	assumeEnabled := flag.Bool("assume-enabled", false, "把当前关闭的支持平台分组按保存配置视为已启用")
+	jsonOut := flag.Bool("json", false, "以 JSON 输出完整报告（默认输出可读表格）")
+	flag.Parse()
+	if *inputPath == "" {
+		fmt.Fprintln(os.Stderr, "usage: profit-preview -input dump.json [-assume-enabled] [-json]")
+		os.Exit(2)
+	}
+	raw, err := os.ReadFile(*inputPath)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "read input: %v\n", err)
+		os.Exit(1)
+	}
+	inputs, err := parsePreviewInputs(raw, *assumeEnabled)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "parse input: %v\n", err)
+		os.Exit(1)
+	}
+
+	evalAt := time.Now()
+	reports := service.PreviewProfitAdmission(inputs, evalAt)
+	if len(reports) == 0 {
+		fmt.Fprintln(os.Stderr, "input produced no preview reports")
+		os.Exit(1)
+	}
+
+	if *jsonOut {
+		enc := json.NewEncoder(os.Stdout)
+		enc.SetIndent("", "  ")
+		if err := enc.Encode(map[string]any{"evaluated_at": evalAt, "reports": reports}); err != nil {
+			fmt.Fprintf(os.Stderr, "write output: %v\n", err)
+			os.Exit(1)
+		}
+		return
+	}
+
+	fmt.Printf("利润门预演 @ %s（U=账号倍率；探测状态仅告警）\n", evalAt.Format(time.RFC3339))
+	for _, report := range reports {
+		fmt.Printf("\n== 分组 %d %s [%s] ==\n", report.GroupID, report.GroupName, report.Platform)
+		fmt.Printf("  利润门生效=%v 假定启用=%v | 默认 D=%.4f 阈值=%.4f | 最低有效 D=%.4f 阈值=%.4f\n",
+			report.EffectiveGate, report.AssumedEnabled,
+			report.DefaultD, report.ThresholdDefault, report.MinEffectiveD, report.ThresholdMinD)
+		counts := map[string]int{}
+		for _, v := range report.Verdicts {
+			counts[v.Class]++
+			rate := "-"
+			if v.AccountRate != nil {
+				rate = fmt.Sprintf("%.4f", *v.AccountRate)
+			}
+			flags := make([]string, 0, 2)
+			if v.RejectedUnderMinD {
+				flags = append(flags, "最低有效D下拒绝")
+			}
+			if len(v.Warnings) > 0 {
+				flags = append(flags, strings.Join(v.Warnings, ","))
+			}
+			suffix := ""
+			if len(flags) > 0 {
+				suffix = "  [" + strings.Join(flags, "; ") + "]"
+			}
+			fmt.Printf("  账号 %-4d %-24s 平台=%-12s U=%-8s 来源=%-19s %s%s\n",
+				v.AccountID, v.Name, v.Platform, rate, v.RateSource, v.Class, suffix)
+		}
+		fmt.Printf("  分类合计: 准入=%d 利润不足=%d 倍率非法=%d\n",
+			counts[service.ProfitPreviewClassAdmitted],
+			counts[service.ProfitPreviewClassRejectedThreshold],
+			counts[service.ProfitPreviewClassRejectedInvalidRate])
+		models := make([]string, 0, len(report.RemainingByModel))
+		for model := range report.RemainingByModel {
+			models = append(models, model)
+		}
+		sort.Strings(models)
+		for _, model := range models {
+			fmt.Printf("  模型 %-20s 利润门准入账号: 默认D=%d 最低有效D=%d\n",
+				model, report.RemainingByModel[model], report.RemainingByModelMinD[model])
+		}
+		for _, model := range modelsWithZeroRemaining(report) {
+			fmt.Printf("  警告: 模型 %s 启用后利润门准入账号为 0\n", model)
+		}
+		for _, model := range modelsWithZeroRemainingUnderMinD(report) {
+			fmt.Printf("  警告: 模型 %s 在最低有效D（存在低倍率用户覆盖）下利润门准入账号为 0\n", model)
+		}
+	}
+}
+
+func parsePreviewInputs(raw []byte, assumeEnabled bool) ([]service.ProfitPreviewGroupInput, error) {
+	var doc inputDoc
+	if err := json.Unmarshal(raw, &doc); err != nil {
+		return nil, err
+	}
+	if len(doc.Groups) == 0 {
+		return nil, fmt.Errorf("input contains no groups; check the export query and target configuration")
+	}
+
+	inputs := make([]service.ProfitPreviewGroupInput, 0, len(doc.Groups))
+	for i, entry := range doc.Groups {
+		if entry.Group.ID <= 0 || strings.TrimSpace(entry.Group.Platform) == "" {
+			return nil, fmt.Errorf("invalid group at index %d: id and platform are required", i)
+		}
+		group := &service.Group{
+			ID:                   entry.Group.ID,
+			Name:                 entry.Group.Name,
+			Platform:             entry.Group.Platform,
+			Status:               service.StatusActive,
+			Hydrated:             true,
+			RateMultiplier:       entry.Group.RateMultiplier,
+			SubscriptionType:     entry.Group.SubscriptionType,
+			ProfitControlEnabled: entry.Group.ProfitControlEnabled,
+			ProfitMinMargin:      entry.Group.ProfitMinMargin,
+			ProfitSafetyBuffer:   entry.Group.ProfitSafetyBuffer,
+			PeakRateEnabled:      entry.Group.PeakRateEnabled,
+			PeakStart:            entry.Group.PeakStart,
+			PeakEnd:              entry.Group.PeakEnd,
+			PeakRateMultiplier:   entry.Group.PeakRateMultiplier,
+		}
+		accounts := make([]*service.Account, 0, len(entry.Accounts))
+		for _, a := range entry.Accounts {
+			account := &service.Account{
+				ID:             a.ID,
+				Name:           a.Name,
+				Platform:       a.Platform,
+				Type:           a.Type,
+				RateMultiplier: a.RateMultiplier,
+				Extra:          a.Extra,
+			}
+			if len(a.ModelMapping) > 0 {
+				mapping := make(map[string]any, len(a.ModelMapping))
+				for k, v := range a.ModelMapping {
+					mapping[k] = v
+				}
+				account.Credentials = map[string]any{"model_mapping": mapping}
+			}
+			accounts = append(accounts, account)
+		}
+		overrides := make(map[int64]float64, len(entry.UserOverrides))
+		for userID, rate := range entry.UserOverrides {
+			if rate == nil {
+				continue
+			}
+			var id int64
+			if _, err := fmt.Sscan(userID, &id); err == nil && id > 0 {
+				overrides[id] = *rate
+			}
+		}
+		inputs = append(inputs, service.ProfitPreviewGroupInput{
+			Group:         group,
+			Accounts:      accounts,
+			UserOverrides: overrides,
+			Models:        entry.Models,
+			AssumeEnabled: assumeEnabled,
+		})
+	}
+	return inputs, nil
+}
+
+func modelsWithZeroRemaining(report service.ProfitPreviewGroupReport) []string {
+	var out []string
+	for model, count := range report.RemainingByModel {
+		if count == 0 {
+			out = append(out, model)
+		}
+	}
+	sort.Strings(out)
+	return out
+}
+
+// modelsWithZeroRemainingUnderMinD 返回默认 D 下仍有准入账号、但在最低有效 D
+// 下会归零的模型。最低有效 D 来自分组内最低的用户级倍率覆盖：这些模型对那部分
+// 用户是全黑的，而只看默认 D 的告警完全看不出来。
+// 两档都为 0 的模型由 modelsWithZeroRemaining 报告，这里不重复。
+func modelsWithZeroRemainingUnderMinD(report service.ProfitPreviewGroupReport) []string {
+	var out []string
+	for model, count := range report.RemainingByModelMinD {
+		if count == 0 && report.RemainingByModel[model] > 0 {
+			out = append(out, model)
+		}
+	}
+	sort.Strings(out)
+	return out
+}

backend/cmd/profit-preview/main_test.go

@@ -0,0 +1,92 @@
+package main
+
+import (
+	"testing"
+	"time"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/stretchr/testify/require"
+)
+
+func TestParsePreviewInputsIgnoresNullUserOverride(t *testing.T) {
+	raw := []byte(`{
+		"groups": [{
+			"group": {
+				"id": 50,
+				"name": "preview",
+				"platform": "openai",
+				"rate_multiplier": 0.5,
+				"subscription_type": "standard",
+				"profit_control_enabled": false,
+				"profit_min_margin": 0.1,
+				"profit_safety_buffer": 0
+			},
+			"accounts": [{
+				"id": 1,
+				"name": "cheap",
+				"platform": "openai",
+				"type": "apikey",
+				"rate_multiplier": 0.2
+			}],
+			"user_overrides": {"40": null, "41": 0.4},
+			"models": ["gpt-test"]
+		}]
+	}`)
+
+	inputs, err := parsePreviewInputs(raw, true)
+	require.NoError(t, err)
+	require.Len(t, inputs, 1)
+	require.Equal(t, map[int64]float64{41: 0.4}, inputs[0].UserOverrides)
+	require.True(t, inputs[0].AssumeEnabled)
+
+	report := service.PreviewProfitAdmission(inputs, time.Date(2026, 1, 15, 8, 30, 0, 0, time.UTC))[0]
+	require.InDelta(t, 0.4, report.MinEffectiveD, 1e-12, "null 覆盖不能被解码成 0 倍率")
+	require.InDelta(t, 0.36, report.ThresholdMinD, 1e-12)
+}
+
+func TestParsePreviewInputsRejectsEmptyGroups(t *testing.T) {
+	for _, raw := range [][]byte{
+		[]byte(`{"groups":null}`),
+		[]byte(`{"groups":[]}`),
+	} {
+		inputs, err := parsePreviewInputs(raw, false)
+		require.ErrorContains(t, err, "input contains no groups")
+		require.Nil(t, inputs)
+	}
+}
+
+// TestModelsWithZeroRemainingWarnings 钉死两档 D 的归零告警分工：
+// 默认 D 归零由 modelsWithZeroRemaining 报告；默认 D 仍有账号但最低有效 D
+// （分组内存在更低的用户级倍率覆盖）归零的模型必须单独告警——那些用户的该
+// 模型会全黑，只看默认 D 完全看不出来。两档都为 0 时不重复告警。
+func TestModelsWithZeroRemainingWarnings(t *testing.T) {
+	report := service.ProfitPreviewGroupReport{
+		RemainingByModel: map[string]int{
+			"both-zero":      0,
+			"min-d-zero":     2,
+			"healthy":        3,
+			"min-d-zero-alt": 1,
+		},
+		RemainingByModelMinD: map[string]int{
+			"both-zero":      0,
+			"min-d-zero":     0,
+			"healthy":        3,
+			"min-d-zero-alt": 0,
+		},
+	}
+
+	if got := modelsWithZeroRemaining(report); len(got) != 1 || got[0] != "both-zero" {
+		t.Fatalf("默认D归零告警应只覆盖 both-zero，got %v", got)
+	}
+
+	got := modelsWithZeroRemainingUnderMinD(report)
+	want := []string{"min-d-zero", "min-d-zero-alt"}
+	if len(got) != len(want) {
+		t.Fatalf("最低有效D归零告警不符: got %v want %v", got, want)
+	}
+	for i := range want {
+		if got[i] != want[i] {
+			t.Fatalf("最低有效D归零告警不符（应按模型名排序）: got %v want %v", got, want)
+		}
+	}
+}

backend/cmd/server/VERSION

@@ -1 +1 @@
-0.1.163
+0.1.169

backend/cmd/server/wire_gen.go

@@ -253,9 +253,9 @@ func initializeApplication(buildInfo handler.BuildInfo) (*Application, error) {
 	channelMonitorRequestTemplateHandler := admin.NewChannelMonitorRequestTemplateHandler(channelMonitorRequestTemplateService)
 	contentModerationRepository := repository.NewContentModerationRepository(db)
 	contentModerationHashCache := repository.NewContentModerationHashCache(redisClient)
-	contentModerationService := service.NewContentModerationService(settingRepository, contentModerationRepository, contentModerationHashCache, groupRepository, userRepository, apiKeyAuthCacheInvalidator, emailService)
+	contentModerationService := service.NewContentModerationService(settingRepository, contentModerationRepository, contentModerationHashCache, groupRepository, userRepository, proxyRepository, apiKeyAuthCacheInvalidator, emailService)
 	contentModerationHandler := admin.NewContentModerationHandler(contentModerationService)
-	configManager := securityaudit.NewConfigManager(db, settingRepository, redisClient, secretEncryptor)
+	configManager := securityaudit.NewConfigManager(db, settingRepository, redisClient, secretEncryptor, configConfig)
 	postgreSQLRepository := securityaudit.NewPostgreSQLRepository(db)
 	redisPayloadStore := securityaudit.NewRedisPayloadStore(redisClient)
 	openAICompatibleScanner := securityaudit.NewOpenAICompatibleScanner()
@@ -280,9 +280,17 @@ func initializeApplication(buildInfo handler.BuildInfo) (*Application, error) {
 	openAIGatewayHandler := handler.ProvideOpenAIGatewayHandler(openAIGatewayService, concurrencyService, billingCacheService, apiKeyService, usageRecordWorkerPool, errorPassthroughService, contentModerationService, opsService, grokQuotaService, configConfig, coordinator)
 	handlerSettingHandler := handler.ProvideSettingHandler(settingService, buildInfo, notificationEmailService)
 	totpHandler := handler.NewTotpHandler(totpService)
+	passkeyRepository := repository.NewPasskeyRepository(db)
+	passkeySessionStore := repository.NewPasskeySessionStore(redisClient)
+	passkeyService, err := service.NewPasskeyService(configConfig, passkeyRepository, passkeySessionStore, userRepository)
+	if err != nil {
+		return nil, err
+	}
+	passkeyHandler := handler.NewPasskeyHandler(passkeyService, authService, settingService)
 	handlerPaymentHandler := handler.NewPaymentHandler(paymentService, paymentConfigService)
 	paymentWebhookHandler := handler.NewPaymentWebhookHandler(paymentService, registry)
 	availableChannelHandler := handler.NewAvailableChannelHandler(channelService, apiKeyService, settingService)
+	modelPlazaHandler := handler.NewModelPlazaHandler(channelService, apiKeyService, settingService)
 	imageTaskStore := repository.NewImageTaskStore(redisClient)
 	imageTaskService := service.ProvideImageTaskService(imageTaskStore, imageStorageSettingService)
 	asyncImageHandler := handler.NewAsyncImageHandler(imageTaskService, openAIGatewayHandler)
@@ -296,13 +304,14 @@ func initializeApplication(buildInfo handler.BuildInfo) (*Application, error) {
 	batchImageHandler := handler.ProvideBatchImageHandler(batchImagePublicService, batchImageDownloadService, batchImageCleanupService, openAIGatewayHandler)
 	idempotencyCoordinator := service.ProvideIdempotencyCoordinator(idempotencyRepository, configConfig)
 	idempotencyCleanupService := service.ProvideIdempotencyCleanupService(idempotencyRepository, configConfig)
-	handlers := handler.ProvideHandlers(authHandler, userHandler, apiKeyHandler, usageHandler, redeemHandler, subscriptionHandler, announcementHandler, channelMonitorUserHandler, adminHandlers, gatewayHandler, openAIGatewayHandler, handlerSettingHandler, totpHandler, handlerPaymentHandler, paymentWebhookHandler, availableChannelHandler, asyncImageHandler, batchImageHandler, idempotencyCoordinator, idempotencyCleanupService)
+	handlers := handler.ProvideHandlers(authHandler, userHandler, apiKeyHandler, usageHandler, redeemHandler, subscriptionHandler, announcementHandler, channelMonitorUserHandler, adminHandlers, gatewayHandler, openAIGatewayHandler, handlerSettingHandler, totpHandler, passkeyHandler, handlerPaymentHandler, paymentWebhookHandler, availableChannelHandler, modelPlazaHandler, asyncImageHandler, batchImageHandler, idempotencyCoordinator, idempotencyCleanupService)
 	jwtAuthMiddleware := middleware.NewJWTAuthMiddleware(authService, userService, settingService, auditLogService)
+	optionalJWTAuthMiddleware := middleware.NewOptionalJWTAuthMiddleware(authService, userService, settingService, auditLogService)
 	adminAuthMiddleware := middleware.NewAdminAuthMiddleware(authService, userService, settingService, auditLogService)
 	apiKeyAuthMiddleware := middleware.NewAPIKeyAuthMiddleware(apiKeyService, subscriptionService, configConfig)
 	auditLogMiddleware := middleware.NewAuditLogMiddleware(auditLogService)
 	stepUpAuthMiddleware := middleware.NewStepUpAuthMiddleware(totpService, userService, settingService)
-	engine := server.ProvideRouter(configConfig, handlers, jwtAuthMiddleware, adminAuthMiddleware, apiKeyAuthMiddleware, auditLogMiddleware, stepUpAuthMiddleware, apiKeyService, subscriptionService, opsService, settingService, compositeRouteResolver, redisClient)
+	engine := server.ProvideRouter(configConfig, handlers, jwtAuthMiddleware, optionalJWTAuthMiddleware, adminAuthMiddleware, apiKeyAuthMiddleware, auditLogMiddleware, stepUpAuthMiddleware, apiKeyService, subscriptionService, opsService, settingService, compositeRouteResolver, redisClient)
 	httpServer := server.ProvideHTTPServer(configConfig, engine)
 	opsMetricsCollector := service.ProvideOpsMetricsCollector(opsRepository, settingRepository, accountRepository, concurrencyService, db, redisClient, configConfig)
 	opsAggregationService := service.ProvideOpsAggregationService(opsRepository, settingRepository, db, redisClient, configConfig)

backend/ent/client.go

@@ -6828,25 +6828,25 @@ type (
 		APIKey, Account, AccountGroup, Announcement, AnnouncementRead, AuthIdentity,
 		AuthIdentityChannel, BatchImageEvent, BatchImageItem, BatchImageJob,
 		ChannelMonitor, ChannelMonitorDailyRollup, ChannelMonitorHistory,
-		ChannelMonitorRequestTemplate, CompositeModelRoute,
-		ErrorPassthroughRule, Group, IdempotencyRecord, IdentityAdoptionDecision,
-		PaymentAuditLog, PaymentOrder, PaymentProviderInstance, PendingAuthSession,
-		PromoCode, PromoCodeUsage, Proxy, RedeemCode, SecuritySecret, Setting,
-		SubscriptionPlan, TLSFingerprintProfile, UsageCleanupTask, UsageLog, User,
-		UserAllowedGroup, UserAttributeDefinition, UserAttributeValue,
-		UserPlatformQuota, UserSubscription []ent.Hook
+		ChannelMonitorRequestTemplate, CompositeModelRoute, ErrorPassthroughRule,
+		Group, IdempotencyRecord, IdentityAdoptionDecision, PaymentAuditLog,
+		PaymentOrder, PaymentProviderInstance, PendingAuthSession, PromoCode,
+		PromoCodeUsage, Proxy, RedeemCode, SecuritySecret, Setting, SubscriptionPlan,
+		TLSFingerprintProfile, UsageCleanupTask, UsageLog, User, UserAllowedGroup,
+		UserAttributeDefinition, UserAttributeValue, UserPlatformQuota,
+		UserSubscription []ent.Hook
 	}
 	inters struct {
 		APIKey, Account, AccountGroup, Announcement, AnnouncementRead, AuthIdentity,
 		AuthIdentityChannel, BatchImageEvent, BatchImageItem, BatchImageJob,
 		ChannelMonitor, ChannelMonitorDailyRollup, ChannelMonitorHistory,
-		ChannelMonitorRequestTemplate, CompositeModelRoute,
-		ErrorPassthroughRule, Group, IdempotencyRecord, IdentityAdoptionDecision,
-		PaymentAuditLog, PaymentOrder, PaymentProviderInstance, PendingAuthSession,
-		PromoCode, PromoCodeUsage, Proxy, RedeemCode, SecuritySecret, Setting,
-		SubscriptionPlan, TLSFingerprintProfile, UsageCleanupTask, UsageLog, User,
-		UserAllowedGroup, UserAttributeDefinition, UserAttributeValue,
-		UserPlatformQuota, UserSubscription []ent.Interceptor
+		ChannelMonitorRequestTemplate, CompositeModelRoute, ErrorPassthroughRule,
+		Group, IdempotencyRecord, IdentityAdoptionDecision, PaymentAuditLog,
+		PaymentOrder, PaymentProviderInstance, PendingAuthSession, PromoCode,
+		PromoCodeUsage, Proxy, RedeemCode, SecuritySecret, Setting, SubscriptionPlan,
+		TLSFingerprintProfile, UsageCleanupTask, UsageLog, User, UserAllowedGroup,
+		UserAttributeDefinition, UserAttributeValue, UserPlatformQuota,
+		UserSubscription []ent.Interceptor
 	}
 )
 

backend/ent/group.go

@@ -105,6 +105,8 @@ type Group struct {
 	SortOrder int `json:"sort_order,omitempty"`
 	// 是否允许 /v1/messages 调度到此 OpenAI 分组
 	AllowMessagesDispatch bool `json:"allow_messages_dispatch,omitempty"`
+	// 是否允许此 OpenAI 分组访问 Live 接口
+	AllowLive bool `json:"allow_live,omitempty"`
 	// 仅允许非 apikey 类型账号关联到此分组
 	RequireOauthOnly bool `json:"require_oauth_only,omitempty"`
 	// 调度时仅允许 privacy 已成功设置的账号
@@ -121,6 +123,12 @@ type Group struct {
 	MaxReasoningEffort string `json:"max_reasoning_effort,omitempty"`
 	// OpenAI reasoning effort 自定义精确映射；先映射再应用上限
 	ReasoningEffortMappings []domain.ReasoningEffortMapping `json:"reasoning_effort_mappings,omitempty"`
+	// 是否启用利润控制：调度时仅允许账号计费倍率满足毛利率要求的账号进入候选池
+	ProfitControlEnabled bool `json:"profit_control_enabled,omitempty"`
+	// 最低毛利率，小数（0.30=30%）；账号准入条件为 U <= D*(1-margin-buffer)
+	ProfitMinMargin float64 `json:"profit_min_margin,omitempty"`
+	// 安全缓冲，小数；与 margin 相加后从下游倍率中扣除，默认 0
+	ProfitSafetyBuffer float64 `json:"profit_safety_buffer,omitempty"`
 	// Edges holds the relations/edges for other nodes in the graph.
 	// The values are being populated by the GroupQuery when eager-loading is set.
 	Edges        GroupEdges `json:"edges"`
@@ -229,9 +237,9 @@ func (*Group) scanValues(columns []string) ([]any, error) {
 		switch columns[i] {
 		case group.FieldModelRouting, group.FieldSupportedModelScopes, group.FieldMessagesDispatchModelConfig, group.FieldModelsListConfig, group.FieldReasoningEffortMappings:
 			values[i] = new([]byte)
-		case group.FieldPeakRateEnabled, group.FieldIsExclusive, group.FieldAllowImageGeneration, group.FieldAllowBatchImageGeneration, group.FieldImageRateIndependent, group.FieldVideoRateIndependent, group.FieldClaudeCodeOnly, group.FieldModelRoutingEnabled, group.FieldMcpXMLInject, group.FieldAllowMessagesDispatch, group.FieldRequireOauthOnly, group.FieldRequirePrivacySet:
+		case group.FieldPeakRateEnabled, group.FieldIsExclusive, group.FieldAllowImageGeneration, group.FieldAllowBatchImageGeneration, group.FieldImageRateIndependent, group.FieldVideoRateIndependent, group.FieldClaudeCodeOnly, group.FieldModelRoutingEnabled, group.FieldMcpXMLInject, group.FieldAllowMessagesDispatch, group.FieldAllowLive, group.FieldRequireOauthOnly, group.FieldRequirePrivacySet, group.FieldProfitControlEnabled:
 			values[i] = new(sql.NullBool)
-		case group.FieldRateMultiplier, group.FieldPeakRateMultiplier, group.FieldDailyLimitUsd, group.FieldWeeklyLimitUsd, group.FieldMonthlyLimitUsd, group.FieldImageRateMultiplier, group.FieldImagePrice1k, group.FieldImagePrice2k, group.FieldImagePrice4k, group.FieldBatchImageDiscountMultiplier, group.FieldBatchImageHoldMultiplier, group.FieldVideoRateMultiplier, group.FieldVideoPrice480p, group.FieldVideoPrice720p, group.FieldVideoPrice1080p, group.FieldWebSearchPricePerCall:
+		case group.FieldRateMultiplier, group.FieldPeakRateMultiplier, group.FieldDailyLimitUsd, group.FieldWeeklyLimitUsd, group.FieldMonthlyLimitUsd, group.FieldImageRateMultiplier, group.FieldImagePrice1k, group.FieldImagePrice2k, group.FieldImagePrice4k, group.FieldBatchImageDiscountMultiplier, group.FieldBatchImageHoldMultiplier, group.FieldVideoRateMultiplier, group.FieldVideoPrice480p, group.FieldVideoPrice720p, group.FieldVideoPrice1080p, group.FieldWebSearchPricePerCall, group.FieldProfitMinMargin, group.FieldProfitSafetyBuffer:
 			values[i] = new(sql.NullFloat64)
 		case group.FieldID, group.FieldDefaultValidityDays, group.FieldFallbackGroupID, group.FieldFallbackGroupIDOnInvalidRequest, group.FieldSortOrder, group.FieldRpmLimit:
 			values[i] = new(sql.NullInt64)
@@ -537,6 +545,12 @@ func (_m *Group) assignValues(columns []string, values []any) error {
 			} else if value.Valid {
 				_m.AllowMessagesDispatch = value.Bool
 			}
+		case group.FieldAllowLive:
+			if value, ok := values[i].(*sql.NullBool); !ok {
+				return fmt.Errorf("unexpected type %T for field allow_live", values[i])
+			} else if value.Valid {
+				_m.AllowLive = value.Bool
+			}
 		case group.FieldRequireOauthOnly:
 			if value, ok := values[i].(*sql.NullBool); !ok {
 				return fmt.Errorf("unexpected type %T for field require_oauth_only", values[i])
@@ -591,6 +605,24 @@ func (_m *Group) assignValues(columns []string, values []any) error {
 					return fmt.Errorf("unmarshal field reasoning_effort_mappings: %w", err)
 				}
 			}
+		case group.FieldProfitControlEnabled:
+			if value, ok := values[i].(*sql.NullBool); !ok {
+				return fmt.Errorf("unexpected type %T for field profit_control_enabled", values[i])
+			} else if value.Valid {
+				_m.ProfitControlEnabled = value.Bool
+			}
+		case group.FieldProfitMinMargin:
+			if value, ok := values[i].(*sql.NullFloat64); !ok {
+				return fmt.Errorf("unexpected type %T for field profit_min_margin", values[i])
+			} else if value.Valid {
+				_m.ProfitMinMargin = value.Float64
+			}
+		case group.FieldProfitSafetyBuffer:
+			if value, ok := values[i].(*sql.NullFloat64); !ok {
+				return fmt.Errorf("unexpected type %T for field profit_safety_buffer", values[i])
+			} else if value.Valid {
+				_m.ProfitSafetyBuffer = value.Float64
+			}
 		default:
 			_m.selectValues.Set(columns[i], values[i])
 		}
@@ -826,6 +858,9 @@ func (_m *Group) String() string {
 	builder.WriteString("allow_messages_dispatch=")
 	builder.WriteString(fmt.Sprintf("%v", _m.AllowMessagesDispatch))
 	builder.WriteString(", ")
+	builder.WriteString("allow_live=")
+	builder.WriteString(fmt.Sprintf("%v", _m.AllowLive))
+	builder.WriteString(", ")
 	builder.WriteString("require_oauth_only=")
 	builder.WriteString(fmt.Sprintf("%v", _m.RequireOauthOnly))
 	builder.WriteString(", ")
@@ -849,6 +884,15 @@ func (_m *Group) String() string {
 	builder.WriteString(", ")
 	builder.WriteString("reasoning_effort_mappings=")
 	builder.WriteString(fmt.Sprintf("%v", _m.ReasoningEffortMappings))
+	builder.WriteString(", ")
+	builder.WriteString("profit_control_enabled=")
+	builder.WriteString(fmt.Sprintf("%v", _m.ProfitControlEnabled))
+	builder.WriteString(", ")
+	builder.WriteString("profit_min_margin=")
+	builder.WriteString(fmt.Sprintf("%v", _m.ProfitMinMargin))
+	builder.WriteString(", ")
+	builder.WriteString("profit_safety_buffer=")
+	builder.WriteString(fmt.Sprintf("%v", _m.ProfitSafetyBuffer))
 	builder.WriteByte(')')
 	return builder.String()
 }

backend/ent/group/group.go

@@ -102,6 +102,8 @@ const (
 	FieldSortOrder = "sort_order"
 	// FieldAllowMessagesDispatch holds the string denoting the allow_messages_dispatch field in the database.
 	FieldAllowMessagesDispatch = "allow_messages_dispatch"
+	// FieldAllowLive holds the string denoting the allow_live field in the database.
+	FieldAllowLive = "allow_live"
 	// FieldRequireOauthOnly holds the string denoting the require_oauth_only field in the database.
 	FieldRequireOauthOnly = "require_oauth_only"
 	// FieldRequirePrivacySet holds the string denoting the require_privacy_set field in the database.
@@ -118,6 +120,12 @@ const (
 	FieldMaxReasoningEffort = "max_reasoning_effort"
 	// FieldReasoningEffortMappings holds the string denoting the reasoning_effort_mappings field in the database.
 	FieldReasoningEffortMappings = "reasoning_effort_mappings"
+	// FieldProfitControlEnabled holds the string denoting the profit_control_enabled field in the database.
+	FieldProfitControlEnabled = "profit_control_enabled"
+	// FieldProfitMinMargin holds the string denoting the profit_min_margin field in the database.
+	FieldProfitMinMargin = "profit_min_margin"
+	// FieldProfitSafetyBuffer holds the string denoting the profit_safety_buffer field in the database.
+	FieldProfitSafetyBuffer = "profit_safety_buffer"
 	// EdgeAPIKeys holds the string denoting the api_keys edge name in mutations.
 	EdgeAPIKeys = "api_keys"
 	// EdgeRedeemCodes holds the string denoting the redeem_codes edge name in mutations.
@@ -236,6 +244,7 @@ var Columns = []string{
 	FieldSupportedModelScopes,
 	FieldSortOrder,
 	FieldAllowMessagesDispatch,
+	FieldAllowLive,
 	FieldRequireOauthOnly,
 	FieldRequirePrivacySet,
 	FieldDefaultMappedModel,
@@ -244,6 +253,9 @@ var Columns = []string{
 	FieldRpmLimit,
 	FieldMaxReasoningEffort,
 	FieldReasoningEffortMappings,
+	FieldProfitControlEnabled,
+	FieldProfitMinMargin,
+	FieldProfitSafetyBuffer,
 }
 
 var (
@@ -341,6 +353,8 @@ var (
 	DefaultSortOrder int
 	// DefaultAllowMessagesDispatch holds the default value on creation for the "allow_messages_dispatch" field.
 	DefaultAllowMessagesDispatch bool
+	// DefaultAllowLive holds the default value on creation for the "allow_live" field.
+	DefaultAllowLive bool
 	// DefaultRequireOauthOnly holds the default value on creation for the "require_oauth_only" field.
 	DefaultRequireOauthOnly bool
 	// DefaultRequirePrivacySet holds the default value on creation for the "require_privacy_set" field.
@@ -361,6 +375,12 @@ var (
 	MaxReasoningEffortValidator func(string) error
 	// DefaultReasoningEffortMappings holds the default value on creation for the "reasoning_effort_mappings" field.
 	DefaultReasoningEffortMappings []domain.ReasoningEffortMapping
+	// DefaultProfitControlEnabled holds the default value on creation for the "profit_control_enabled" field.
+	DefaultProfitControlEnabled bool
+	// DefaultProfitMinMargin holds the default value on creation for the "profit_min_margin" field.
+	DefaultProfitMinMargin float64
+	// DefaultProfitSafetyBuffer holds the default value on creation for the "profit_safety_buffer" field.
+	DefaultProfitSafetyBuffer float64
 )
 
 // OrderOption defines the ordering options for the Group queries.
@@ -576,6 +596,11 @@ func ByAllowMessagesDispatch(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldAllowMessagesDispatch, opts...).ToFunc()
 }
 
+// ByAllowLive orders the results by the allow_live field.
+func ByAllowLive(opts ...sql.OrderTermOption) OrderOption {
+	return sql.OrderByField(FieldAllowLive, opts...).ToFunc()
+}
+
 // ByRequireOauthOnly orders the results by the require_oauth_only field.
 func ByRequireOauthOnly(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldRequireOauthOnly, opts...).ToFunc()
@@ -601,6 +626,21 @@ func ByMaxReasoningEffort(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldMaxReasoningEffort, opts...).ToFunc()
 }
 
+// ByProfitControlEnabled orders the results by the profit_control_enabled field.
+func ByProfitControlEnabled(opts ...sql.OrderTermOption) OrderOption {
+	return sql.OrderByField(FieldProfitControlEnabled, opts...).ToFunc()
+}
+
+// ByProfitMinMargin orders the results by the profit_min_margin field.
+func ByProfitMinMargin(opts ...sql.OrderTermOption) OrderOption {
+	return sql.OrderByField(FieldProfitMinMargin, opts...).ToFunc()
+}
+
+// ByProfitSafetyBuffer orders the results by the profit_safety_buffer field.
+func ByProfitSafetyBuffer(opts ...sql.OrderTermOption) OrderOption {
+	return sql.OrderByField(FieldProfitSafetyBuffer, opts...).ToFunc()
+}
+
 // ByAPIKeysCount orders the results by api_keys count.
 func ByAPIKeysCount(opts ...sql.OrderTermOption) OrderOption {
 	return func(s *sql.Selector) {

backend/ent/group/where.go

@@ -260,6 +260,11 @@ func AllowMessagesDispatch(v bool) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldAllowMessagesDispatch, v))
 }
 
+// AllowLive applies equality check predicate on the "allow_live" field. It's identical to AllowLiveEQ.
+func AllowLive(v bool) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldAllowLive, v))
+}
+
 // RequireOauthOnly applies equality check predicate on the "require_oauth_only" field. It's identical to RequireOauthOnlyEQ.
 func RequireOauthOnly(v bool) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldRequireOauthOnly, v))
@@ -285,6 +290,21 @@ func MaxReasoningEffort(v string) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldMaxReasoningEffort, v))
 }
 
+// ProfitControlEnabled applies equality check predicate on the "profit_control_enabled" field. It's identical to ProfitControlEnabledEQ.
+func ProfitControlEnabled(v bool) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldProfitControlEnabled, v))
+}
+
+// ProfitMinMargin applies equality check predicate on the "profit_min_margin" field. It's identical to ProfitMinMarginEQ.
+func ProfitMinMargin(v float64) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldProfitMinMargin, v))
+}
+
+// ProfitSafetyBuffer applies equality check predicate on the "profit_safety_buffer" field. It's identical to ProfitSafetyBufferEQ.
+func ProfitSafetyBuffer(v float64) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldProfitSafetyBuffer, v))
+}
+
 // CreatedAtEQ applies the EQ predicate on the "created_at" field.
 func CreatedAtEQ(v time.Time) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldCreatedAt, v))
@@ -1985,6 +2005,16 @@ func AllowMessagesDispatchNEQ(v bool) predicate.Group {
 	return predicate.Group(sql.FieldNEQ(FieldAllowMessagesDispatch, v))
 }
 
+// AllowLiveEQ applies the EQ predicate on the "allow_live" field.
+func AllowLiveEQ(v bool) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldAllowLive, v))
+}
+
+// AllowLiveNEQ applies the NEQ predicate on the "allow_live" field.
+func AllowLiveNEQ(v bool) predicate.Group {
+	return predicate.Group(sql.FieldNEQ(FieldAllowLive, v))
+}
+
 // RequireOauthOnlyEQ applies the EQ predicate on the "require_oauth_only" field.
 func RequireOauthOnlyEQ(v bool) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldRequireOauthOnly, v))
@@ -2175,6 +2205,96 @@ func MaxReasoningEffortContainsFold(v string) predicate.Group {
 	return predicate.Group(sql.FieldContainsFold(FieldMaxReasoningEffort, v))
 }
 
+// ProfitControlEnabledEQ applies the EQ predicate on the "profit_control_enabled" field.
+func ProfitControlEnabledEQ(v bool) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldProfitControlEnabled, v))
+}
+
+// ProfitControlEnabledNEQ applies the NEQ predicate on the "profit_control_enabled" field.
+func ProfitControlEnabledNEQ(v bool) predicate.Group {
+	return predicate.Group(sql.FieldNEQ(FieldProfitControlEnabled, v))
+}
+
+// ProfitMinMarginEQ applies the EQ predicate on the "profit_min_margin" field.
+func ProfitMinMarginEQ(v float64) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldProfitMinMargin, v))
+}
+
+// ProfitMinMarginNEQ applies the NEQ predicate on the "profit_min_margin" field.
+func ProfitMinMarginNEQ(v float64) predicate.Group {
+	return predicate.Group(sql.FieldNEQ(FieldProfitMinMargin, v))
+}
+
+// ProfitMinMarginIn applies the In predicate on the "profit_min_margin" field.
+func ProfitMinMarginIn(vs ...float64) predicate.Group {
+	return predicate.Group(sql.FieldIn(FieldProfitMinMargin, vs...))
+}
+
+// ProfitMinMarginNotIn applies the NotIn predicate on the "profit_min_margin" field.
+func ProfitMinMarginNotIn(vs ...float64) predicate.Group {
+	return predicate.Group(sql.FieldNotIn(FieldProfitMinMargin, vs...))
+}
+
+// ProfitMinMarginGT applies the GT predicate on the "profit_min_margin" field.
+func ProfitMinMarginGT(v float64) predicate.Group {
+	return predicate.Group(sql.FieldGT(FieldProfitMinMargin, v))
+}
+
+// ProfitMinMarginGTE applies the GTE predicate on the "profit_min_margin" field.
+func ProfitMinMarginGTE(v float64) predicate.Group {
+	return predicate.Group(sql.FieldGTE(FieldProfitMinMargin, v))
+}
+
+// ProfitMinMarginLT applies the LT predicate on the "profit_min_margin" field.
+func ProfitMinMarginLT(v float64) predicate.Group {
+	return predicate.Group(sql.FieldLT(FieldProfitMinMargin, v))
+}
+
+// ProfitMinMarginLTE applies the LTE predicate on the "profit_min_margin" field.
+func ProfitMinMarginLTE(v float64) predicate.Group {
+	return predicate.Group(sql.FieldLTE(FieldProfitMinMargin, v))
+}
+
+// ProfitSafetyBufferEQ applies the EQ predicate on the "profit_safety_buffer" field.
+func ProfitSafetyBufferEQ(v float64) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldProfitSafetyBuffer, v))
+}
+
+// ProfitSafetyBufferNEQ applies the NEQ predicate on the "profit_safety_buffer" field.
+func ProfitSafetyBufferNEQ(v float64) predicate.Group {
+	return predicate.Group(sql.FieldNEQ(FieldProfitSafetyBuffer, v))
+}
+
+// ProfitSafetyBufferIn applies the In predicate on the "profit_safety_buffer" field.
+func ProfitSafetyBufferIn(vs ...float64) predicate.Group {
+	return predicate.Group(sql.FieldIn(FieldProfitSafetyBuffer, vs...))
+}
+
+// ProfitSafetyBufferNotIn applies the NotIn predicate on the "profit_safety_buffer" field.
+func ProfitSafetyBufferNotIn(vs ...float64) predicate.Group {
+	return predicate.Group(sql.FieldNotIn(FieldProfitSafetyBuffer, vs...))
+}
+
+// ProfitSafetyBufferGT applies the GT predicate on the "profit_safety_buffer" field.
+func ProfitSafetyBufferGT(v float64) predicate.Group {
+	return predicate.Group(sql.FieldGT(FieldProfitSafetyBuffer, v))
+}
+
+// ProfitSafetyBufferGTE applies the GTE predicate on the "profit_safety_buffer" field.
+func ProfitSafetyBufferGTE(v float64) predicate.Group {
+	return predicate.Group(sql.FieldGTE(FieldProfitSafetyBuffer, v))
+}
+
+// ProfitSafetyBufferLT applies the LT predicate on the "profit_safety_buffer" field.
+func ProfitSafetyBufferLT(v float64) predicate.Group {
+	return predicate.Group(sql.FieldLT(FieldProfitSafetyBuffer, v))
+}
+
+// ProfitSafetyBufferLTE applies the LTE predicate on the "profit_safety_buffer" field.
+func ProfitSafetyBufferLTE(v float64) predicate.Group {
+	return predicate.Group(sql.FieldLTE(FieldProfitSafetyBuffer, v))
+}
+
 // HasAPIKeys applies the HasEdge predicate on the "api_keys" edge.
 func HasAPIKeys() predicate.Group {
 	return predicate.Group(func(s *sql.Selector) {

backend/ent/group_create.go

@@ -607,6 +607,20 @@ func (_c *GroupCreate) SetNillableAllowMessagesDispatch(v *bool) *GroupCreate {
 	return _c
 }
 
+// SetAllowLive sets the "allow_live" field.
+func (_c *GroupCreate) SetAllowLive(v bool) *GroupCreate {
+	_c.mutation.SetAllowLive(v)
+	return _c
+}
+
+// SetNillableAllowLive sets the "allow_live" field if the given value is not nil.
+func (_c *GroupCreate) SetNillableAllowLive(v *bool) *GroupCreate {
+	if v != nil {
+		_c.SetAllowLive(*v)
+	}
+	return _c
+}
+
 // SetRequireOauthOnly sets the "require_oauth_only" field.
 func (_c *GroupCreate) SetRequireOauthOnly(v bool) *GroupCreate {
 	_c.mutation.SetRequireOauthOnly(v)
@@ -711,6 +725,48 @@ func (_c *GroupCreate) SetReasoningEffortMappings(v []domain.ReasoningEffortMapp
 	return _c
 }
 
+// SetProfitControlEnabled sets the "profit_control_enabled" field.
+func (_c *GroupCreate) SetProfitControlEnabled(v bool) *GroupCreate {
+	_c.mutation.SetProfitControlEnabled(v)
+	return _c
+}
+
+// SetNillableProfitControlEnabled sets the "profit_control_enabled" field if the given value is not nil.
+func (_c *GroupCreate) SetNillableProfitControlEnabled(v *bool) *GroupCreate {
+	if v != nil {
+		_c.SetProfitControlEnabled(*v)
+	}
+	return _c
+}
+
+// SetProfitMinMargin sets the "profit_min_margin" field.
+func (_c *GroupCreate) SetProfitMinMargin(v float64) *GroupCreate {
+	_c.mutation.SetProfitMinMargin(v)
+	return _c
+}
+
+// SetNillableProfitMinMargin sets the "profit_min_margin" field if the given value is not nil.
+func (_c *GroupCreate) SetNillableProfitMinMargin(v *float64) *GroupCreate {
+	if v != nil {
+		_c.SetProfitMinMargin(*v)
+	}
+	return _c
+}
+
+// SetProfitSafetyBuffer sets the "profit_safety_buffer" field.
+func (_c *GroupCreate) SetProfitSafetyBuffer(v float64) *GroupCreate {
+	_c.mutation.SetProfitSafetyBuffer(v)
+	return _c
+}
+
+// SetNillableProfitSafetyBuffer sets the "profit_safety_buffer" field if the given value is not nil.
+func (_c *GroupCreate) SetNillableProfitSafetyBuffer(v *float64) *GroupCreate {
+	if v != nil {
+		_c.SetProfitSafetyBuffer(*v)
+	}
+	return _c
+}
+
 // AddAPIKeyIDs adds the "api_keys" edge to the APIKey entity by IDs.
 func (_c *GroupCreate) AddAPIKeyIDs(ids ...int64) *GroupCreate {
 	_c.mutation.AddAPIKeyIDs(ids...)
@@ -948,6 +1004,10 @@ func (_c *GroupCreate) defaults() error {
 		v := group.DefaultAllowMessagesDispatch
 		_c.mutation.SetAllowMessagesDispatch(v)
 	}
+	if _, ok := _c.mutation.AllowLive(); !ok {
+		v := group.DefaultAllowLive
+		_c.mutation.SetAllowLive(v)
+	}
 	if _, ok := _c.mutation.RequireOauthOnly(); !ok {
 		v := group.DefaultRequireOauthOnly
 		_c.mutation.SetRequireOauthOnly(v)
@@ -980,6 +1040,18 @@ func (_c *GroupCreate) defaults() error {
 		v := group.DefaultReasoningEffortMappings
 		_c.mutation.SetReasoningEffortMappings(v)
 	}
+	if _, ok := _c.mutation.ProfitControlEnabled(); !ok {
+		v := group.DefaultProfitControlEnabled
+		_c.mutation.SetProfitControlEnabled(v)
+	}
+	if _, ok := _c.mutation.ProfitMinMargin(); !ok {
+		v := group.DefaultProfitMinMargin
+		_c.mutation.SetProfitMinMargin(v)
+	}
+	if _, ok := _c.mutation.ProfitSafetyBuffer(); !ok {
+		v := group.DefaultProfitSafetyBuffer
+		_c.mutation.SetProfitSafetyBuffer(v)
+	}
 	return nil
 }
 
@@ -1101,6 +1173,9 @@ func (_c *GroupCreate) check() error {
 	if _, ok := _c.mutation.AllowMessagesDispatch(); !ok {
 		return &ValidationError{Name: "allow_messages_dispatch", err: errors.New(`ent: missing required field "Group.allow_messages_dispatch"`)}
 	}
+	if _, ok := _c.mutation.AllowLive(); !ok {
+		return &ValidationError{Name: "allow_live", err: errors.New(`ent: missing required field "Group.allow_live"`)}
+	}
 	if _, ok := _c.mutation.RequireOauthOnly(); !ok {
 		return &ValidationError{Name: "require_oauth_only", err: errors.New(`ent: missing required field "Group.require_oauth_only"`)}
 	}
@@ -1135,6 +1210,15 @@ func (_c *GroupCreate) check() error {
 	if _, ok := _c.mutation.ReasoningEffortMappings(); !ok {
 		return &ValidationError{Name: "reasoning_effort_mappings", err: errors.New(`ent: missing required field "Group.reasoning_effort_mappings"`)}
 	}
+	if _, ok := _c.mutation.ProfitControlEnabled(); !ok {
+		return &ValidationError{Name: "profit_control_enabled", err: errors.New(`ent: missing required field "Group.profit_control_enabled"`)}
+	}
+	if _, ok := _c.mutation.ProfitMinMargin(); !ok {
+		return &ValidationError{Name: "profit_min_margin", err: errors.New(`ent: missing required field "Group.profit_min_margin"`)}
+	}
+	if _, ok := _c.mutation.ProfitSafetyBuffer(); !ok {
+		return &ValidationError{Name: "profit_safety_buffer", err: errors.New(`ent: missing required field "Group.profit_safety_buffer"`)}
+	}
 	return nil
 }
 
@@ -1334,6 +1418,10 @@ func (_c *GroupCreate) createSpec() (*Group, *sqlgraph.CreateSpec) {
 		_spec.SetField(group.FieldAllowMessagesDispatch, field.TypeBool, value)
 		_node.AllowMessagesDispatch = value
 	}
+	if value, ok := _c.mutation.AllowLive(); ok {
+		_spec.SetField(group.FieldAllowLive, field.TypeBool, value)
+		_node.AllowLive = value
+	}
 	if value, ok := _c.mutation.RequireOauthOnly(); ok {
 		_spec.SetField(group.FieldRequireOauthOnly, field.TypeBool, value)
 		_node.RequireOauthOnly = value
@@ -1366,6 +1454,18 @@ func (_c *GroupCreate) createSpec() (*Group, *sqlgraph.CreateSpec) {
 		_spec.SetField(group.FieldReasoningEffortMappings, field.TypeJSON, value)
 		_node.ReasoningEffortMappings = value
 	}
+	if value, ok := _c.mutation.ProfitControlEnabled(); ok {
+		_spec.SetField(group.FieldProfitControlEnabled, field.TypeBool, value)
+		_node.ProfitControlEnabled = value
+	}
+	if value, ok := _c.mutation.ProfitMinMargin(); ok {
+		_spec.SetField(group.FieldProfitMinMargin, field.TypeFloat64, value)
+		_node.ProfitMinMargin = value
+	}
+	if value, ok := _c.mutation.ProfitSafetyBuffer(); ok {
+		_spec.SetField(group.FieldProfitSafetyBuffer, field.TypeFloat64, value)
+		_node.ProfitSafetyBuffer = value
+	}
 	if nodes := _c.mutation.APIKeysIDs(); len(nodes) > 0 {
 		edge := &sqlgraph.EdgeSpec{
 			Rel:     sqlgraph.O2M,
@@ -2224,6 +2324,18 @@ func (u *GroupUpsert) UpdateAllowMessagesDispatch() *GroupUpsert {
 	return u
 }
 
+// SetAllowLive sets the "allow_live" field.
+func (u *GroupUpsert) SetAllowLive(v bool) *GroupUpsert {
+	u.Set(group.FieldAllowLive, v)
+	return u
+}
+
+// UpdateAllowLive sets the "allow_live" field to the value that was provided on create.
+func (u *GroupUpsert) UpdateAllowLive() *GroupUpsert {
+	u.SetExcluded(group.FieldAllowLive)
+	return u
+}
+
 // SetRequireOauthOnly sets the "require_oauth_only" field.
 func (u *GroupUpsert) SetRequireOauthOnly(v bool) *GroupUpsert {
 	u.Set(group.FieldRequireOauthOnly, v)
@@ -2326,6 +2438,54 @@ func (u *GroupUpsert) UpdateReasoningEffortMappings() *GroupUpsert {
 	return u
 }
 
+// SetProfitControlEnabled sets the "profit_control_enabled" field.
+func (u *GroupUpsert) SetProfitControlEnabled(v bool) *GroupUpsert {
+	u.Set(group.FieldProfitControlEnabled, v)
+	return u
+}
+
+// UpdateProfitControlEnabled sets the "profit_control_enabled" field to the value that was provided on create.
+func (u *GroupUpsert) UpdateProfitControlEnabled() *GroupUpsert {
+	u.SetExcluded(group.FieldProfitControlEnabled)
+	return u
+}
+
+// SetProfitMinMargin sets the "profit_min_margin" field.
+func (u *GroupUpsert) SetProfitMinMargin(v float64) *GroupUpsert {
+	u.Set(group.FieldProfitMinMargin, v)
+	return u
+}
+
+// UpdateProfitMinMargin sets the "profit_min_margin" field to the value that was provided on create.
+func (u *GroupUpsert) UpdateProfitMinMargin() *GroupUpsert {
+	u.SetExcluded(group.FieldProfitMinMargin)
+	return u
+}
+
+// AddProfitMinMargin adds v to the "profit_min_margin" field.
+func (u *GroupUpsert) AddProfitMinMargin(v float64) *GroupUpsert {
+	u.Add(group.FieldProfitMinMargin, v)
+	return u
+}
+
+// SetProfitSafetyBuffer sets the "profit_safety_buffer" field.
+func (u *GroupUpsert) SetProfitSafetyBuffer(v float64) *GroupUpsert {
+	u.Set(group.FieldProfitSafetyBuffer, v)
+	return u
+}
+
+// UpdateProfitSafetyBuffer sets the "profit_safety_buffer" field to the value that was provided on create.
+func (u *GroupUpsert) UpdateProfitSafetyBuffer() *GroupUpsert {
+	u.SetExcluded(group.FieldProfitSafetyBuffer)
+	return u
+}
+
+// AddProfitSafetyBuffer adds v to the "profit_safety_buffer" field.
+func (u *GroupUpsert) AddProfitSafetyBuffer(v float64) *GroupUpsert {
+	u.Add(group.FieldProfitSafetyBuffer, v)
+	return u
+}
+
 // UpdateNewValues updates the mutable fields using the new values that were set on create.
 // Using this option is equivalent to using:
 //
@@ -3193,6 +3353,20 @@ func (u *GroupUpsertOne) UpdateAllowMessagesDispatch() *GroupUpsertOne {
 	})
 }
 
+// SetAllowLive sets the "allow_live" field.
+func (u *GroupUpsertOne) SetAllowLive(v bool) *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetAllowLive(v)
+	})
+}
+
+// UpdateAllowLive sets the "allow_live" field to the value that was provided on create.
+func (u *GroupUpsertOne) UpdateAllowLive() *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateAllowLive()
+	})
+}
+
 // SetRequireOauthOnly sets the "require_oauth_only" field.
 func (u *GroupUpsertOne) SetRequireOauthOnly(v bool) *GroupUpsertOne {
 	return u.Update(func(s *GroupUpsert) {
@@ -3312,6 +3486,62 @@ func (u *GroupUpsertOne) UpdateReasoningEffortMappings() *GroupUpsertOne {
 	})
 }
 
+// SetProfitControlEnabled sets the "profit_control_enabled" field.
+func (u *GroupUpsertOne) SetProfitControlEnabled(v bool) *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetProfitControlEnabled(v)
+	})
+}
+
+// UpdateProfitControlEnabled sets the "profit_control_enabled" field to the value that was provided on create.
+func (u *GroupUpsertOne) UpdateProfitControlEnabled() *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateProfitControlEnabled()
+	})
+}
+
+// SetProfitMinMargin sets the "profit_min_margin" field.
+func (u *GroupUpsertOne) SetProfitMinMargin(v float64) *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetProfitMinMargin(v)
+	})
+}
+
+// AddProfitMinMargin adds v to the "profit_min_margin" field.
+func (u *GroupUpsertOne) AddProfitMinMargin(v float64) *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.AddProfitMinMargin(v)
+	})
+}
+
+// UpdateProfitMinMargin sets the "profit_min_margin" field to the value that was provided on create.
+func (u *GroupUpsertOne) UpdateProfitMinMargin() *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateProfitMinMargin()
+	})
+}
+
+// SetProfitSafetyBuffer sets the "profit_safety_buffer" field.
+func (u *GroupUpsertOne) SetProfitSafetyBuffer(v float64) *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetProfitSafetyBuffer(v)
+	})
+}
+
+// AddProfitSafetyBuffer adds v to the "profit_safety_buffer" field.
+func (u *GroupUpsertOne) AddProfitSafetyBuffer(v float64) *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.AddProfitSafetyBuffer(v)
+	})
+}
+
+// UpdateProfitSafetyBuffer sets the "profit_safety_buffer" field to the value that was provided on create.
+func (u *GroupUpsertOne) UpdateProfitSafetyBuffer() *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateProfitSafetyBuffer()
+	})
+}
+
 // Exec executes the query.
 func (u *GroupUpsertOne) Exec(ctx context.Context) error {
 	if len(u.create.conflict) == 0 {
@@ -4345,6 +4575,20 @@ func (u *GroupUpsertBulk) UpdateAllowMessagesDispatch() *GroupUpsertBulk {
 	})
 }
 
+// SetAllowLive sets the "allow_live" field.
+func (u *GroupUpsertBulk) SetAllowLive(v bool) *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetAllowLive(v)
+	})
+}
+
+// UpdateAllowLive sets the "allow_live" field to the value that was provided on create.
+func (u *GroupUpsertBulk) UpdateAllowLive() *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateAllowLive()
+	})
+}
+
 // SetRequireOauthOnly sets the "require_oauth_only" field.
 func (u *GroupUpsertBulk) SetRequireOauthOnly(v bool) *GroupUpsertBulk {
 	return u.Update(func(s *GroupUpsert) {
@@ -4464,6 +4708,62 @@ func (u *GroupUpsertBulk) UpdateReasoningEffortMappings() *GroupUpsertBulk {
 	})
 }
 
+// SetProfitControlEnabled sets the "profit_control_enabled" field.
+func (u *GroupUpsertBulk) SetProfitControlEnabled(v bool) *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetProfitControlEnabled(v)
+	})
+}
+
+// UpdateProfitControlEnabled sets the "profit_control_enabled" field to the value that was provided on create.
+func (u *GroupUpsertBulk) UpdateProfitControlEnabled() *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateProfitControlEnabled()
+	})
+}
+
+// SetProfitMinMargin sets the "profit_min_margin" field.
+func (u *GroupUpsertBulk) SetProfitMinMargin(v float64) *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetProfitMinMargin(v)
+	})
+}
+
+// AddProfitMinMargin adds v to the "profit_min_margin" field.
+func (u *GroupUpsertBulk) AddProfitMinMargin(v float64) *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.AddProfitMinMargin(v)
+	})
+}
+
+// UpdateProfitMinMargin sets the "profit_min_margin" field to the value that was provided on create.
+func (u *GroupUpsertBulk) UpdateProfitMinMargin() *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateProfitMinMargin()
+	})
+}
+
+// SetProfitSafetyBuffer sets the "profit_safety_buffer" field.
+func (u *GroupUpsertBulk) SetProfitSafetyBuffer(v float64) *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetProfitSafetyBuffer(v)
+	})
+}
+
+// AddProfitSafetyBuffer adds v to the "profit_safety_buffer" field.
+func (u *GroupUpsertBulk) AddProfitSafetyBuffer(v float64) *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.AddProfitSafetyBuffer(v)
+	})
+}
+
+// UpdateProfitSafetyBuffer sets the "profit_safety_buffer" field to the value that was provided on create.
+func (u *GroupUpsertBulk) UpdateProfitSafetyBuffer() *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateProfitSafetyBuffer()
+	})
+}
+
 // Exec executes the query.
 func (u *GroupUpsertBulk) Exec(ctx context.Context) error {
 	if u.create.err != nil {

backend/ent/group_update.go

@@ -822,6 +822,20 @@ func (_u *GroupUpdate) SetNillableAllowMessagesDispatch(v *bool) *GroupUpdate {
 	return _u
 }
 
+// SetAllowLive sets the "allow_live" field.
+func (_u *GroupUpdate) SetAllowLive(v bool) *GroupUpdate {
+	_u.mutation.SetAllowLive(v)
+	return _u
+}
+
+// SetNillableAllowLive sets the "allow_live" field if the given value is not nil.
+func (_u *GroupUpdate) SetNillableAllowLive(v *bool) *GroupUpdate {
+	if v != nil {
+		_u.SetAllowLive(*v)
+	}
+	return _u
+}
+
 // SetRequireOauthOnly sets the "require_oauth_only" field.
 func (_u *GroupUpdate) SetRequireOauthOnly(v bool) *GroupUpdate {
 	_u.mutation.SetRequireOauthOnly(v)
@@ -939,6 +953,62 @@ func (_u *GroupUpdate) AppendReasoningEffortMappings(v []domain.ReasoningEffortM
 	return _u
 }
 
+// SetProfitControlEnabled sets the "profit_control_enabled" field.
+func (_u *GroupUpdate) SetProfitControlEnabled(v bool) *GroupUpdate {
+	_u.mutation.SetProfitControlEnabled(v)
+	return _u
+}
+
+// SetNillableProfitControlEnabled sets the "profit_control_enabled" field if the given value is not nil.
+func (_u *GroupUpdate) SetNillableProfitControlEnabled(v *bool) *GroupUpdate {
+	if v != nil {
+		_u.SetProfitControlEnabled(*v)
+	}
+	return _u
+}
+
+// SetProfitMinMargin sets the "profit_min_margin" field.
+func (_u *GroupUpdate) SetProfitMinMargin(v float64) *GroupUpdate {
+	_u.mutation.ResetProfitMinMargin()
+	_u.mutation.SetProfitMinMargin(v)
+	return _u
+}
+
+// SetNillableProfitMinMargin sets the "profit_min_margin" field if the given value is not nil.
+func (_u *GroupUpdate) SetNillableProfitMinMargin(v *float64) *GroupUpdate {
+	if v != nil {
+		_u.SetProfitMinMargin(*v)
+	}
+	return _u
+}
+
+// AddProfitMinMargin adds value to the "profit_min_margin" field.
+func (_u *GroupUpdate) AddProfitMinMargin(v float64) *GroupUpdate {
+	_u.mutation.AddProfitMinMargin(v)
+	return _u
+}
+
+// SetProfitSafetyBuffer sets the "profit_safety_buffer" field.
+func (_u *GroupUpdate) SetProfitSafetyBuffer(v float64) *GroupUpdate {
+	_u.mutation.ResetProfitSafetyBuffer()
+	_u.mutation.SetProfitSafetyBuffer(v)
+	return _u
+}
+
+// SetNillableProfitSafetyBuffer sets the "profit_safety_buffer" field if the given value is not nil.
+func (_u *GroupUpdate) SetNillableProfitSafetyBuffer(v *float64) *GroupUpdate {
+	if v != nil {
+		_u.SetProfitSafetyBuffer(*v)
+	}
+	return _u
+}
+
+// AddProfitSafetyBuffer adds value to the "profit_safety_buffer" field.
+func (_u *GroupUpdate) AddProfitSafetyBuffer(v float64) *GroupUpdate {
+	_u.mutation.AddProfitSafetyBuffer(v)
+	return _u
+}
+
 // AddAPIKeyIDs adds the "api_keys" edge to the APIKey entity by IDs.
 func (_u *GroupUpdate) AddAPIKeyIDs(ids ...int64) *GroupUpdate {
 	_u.mutation.AddAPIKeyIDs(ids...)
@@ -1495,6 +1565,9 @@ func (_u *GroupUpdate) sqlSave(ctx context.Context) (_node int, err error) {
 	if value, ok := _u.mutation.AllowMessagesDispatch(); ok {
 		_spec.SetField(group.FieldAllowMessagesDispatch, field.TypeBool, value)
 	}
+	if value, ok := _u.mutation.AllowLive(); ok {
+		_spec.SetField(group.FieldAllowLive, field.TypeBool, value)
+	}
 	if value, ok := _u.mutation.RequireOauthOnly(); ok {
 		_spec.SetField(group.FieldRequireOauthOnly, field.TypeBool, value)
 	}
@@ -1527,6 +1600,21 @@ func (_u *GroupUpdate) sqlSave(ctx context.Context) (_node int, err error) {
 			sqljson.Append(u, group.FieldReasoningEffortMappings, value)
 		})
 	}
+	if value, ok := _u.mutation.ProfitControlEnabled(); ok {
+		_spec.SetField(group.FieldProfitControlEnabled, field.TypeBool, value)
+	}
+	if value, ok := _u.mutation.ProfitMinMargin(); ok {
+		_spec.SetField(group.FieldProfitMinMargin, field.TypeFloat64, value)
+	}
+	if value, ok := _u.mutation.AddedProfitMinMargin(); ok {
+		_spec.AddField(group.FieldProfitMinMargin, field.TypeFloat64, value)
+	}
+	if value, ok := _u.mutation.ProfitSafetyBuffer(); ok {
+		_spec.SetField(group.FieldProfitSafetyBuffer, field.TypeFloat64, value)
+	}
+	if value, ok := _u.mutation.AddedProfitSafetyBuffer(); ok {
+		_spec.AddField(group.FieldProfitSafetyBuffer, field.TypeFloat64, value)
+	}
 	if _u.mutation.APIKeysCleared() {
 		edge := &sqlgraph.EdgeSpec{
 			Rel:     sqlgraph.O2M,
@@ -2627,6 +2715,20 @@ func (_u *GroupUpdateOne) SetNillableAllowMessagesDispatch(v *bool) *GroupUpdate
 	return _u
 }
 
+// SetAllowLive sets the "allow_live" field.
+func (_u *GroupUpdateOne) SetAllowLive(v bool) *GroupUpdateOne {
+	_u.mutation.SetAllowLive(v)
+	return _u
+}
+
+// SetNillableAllowLive sets the "allow_live" field if the given value is not nil.
+func (_u *GroupUpdateOne) SetNillableAllowLive(v *bool) *GroupUpdateOne {
+	if v != nil {
+		_u.SetAllowLive(*v)
+	}
+	return _u
+}
+
 // SetRequireOauthOnly sets the "require_oauth_only" field.
 func (_u *GroupUpdateOne) SetRequireOauthOnly(v bool) *GroupUpdateOne {
 	_u.mutation.SetRequireOauthOnly(v)
@@ -2744,6 +2846,62 @@ func (_u *GroupUpdateOne) AppendReasoningEffortMappings(v []domain.ReasoningEffo
 	return _u
 }
 
+// SetProfitControlEnabled sets the "profit_control_enabled" field.
+func (_u *GroupUpdateOne) SetProfitControlEnabled(v bool) *GroupUpdateOne {
+	_u.mutation.SetProfitControlEnabled(v)
+	return _u
+}
+
+// SetNillableProfitControlEnabled sets the "profit_control_enabled" field if the given value is not nil.
+func (_u *GroupUpdateOne) SetNillableProfitControlEnabled(v *bool) *GroupUpdateOne {
+	if v != nil {
+		_u.SetProfitControlEnabled(*v)
+	}
+	return _u
+}
+
+// SetProfitMinMargin sets the "profit_min_margin" field.
+func (_u *GroupUpdateOne) SetProfitMinMargin(v float64) *GroupUpdateOne {
+	_u.mutation.ResetProfitMinMargin()
+	_u.mutation.SetProfitMinMargin(v)
+	return _u
+}
+
+// SetNillableProfitMinMargin sets the "profit_min_margin" field if the given value is not nil.
+func (_u *GroupUpdateOne) SetNillableProfitMinMargin(v *float64) *GroupUpdateOne {
+	if v != nil {
+		_u.SetProfitMinMargin(*v)
+	}
+	return _u
+}
+
+// AddProfitMinMargin adds value to the "profit_min_margin" field.
+func (_u *GroupUpdateOne) AddProfitMinMargin(v float64) *GroupUpdateOne {
+	_u.mutation.AddProfitMinMargin(v)
+	return _u
+}
+
+// SetProfitSafetyBuffer sets the "profit_safety_buffer" field.
+func (_u *GroupUpdateOne) SetProfitSafetyBuffer(v float64) *GroupUpdateOne {
+	_u.mutation.ResetProfitSafetyBuffer()
+	_u.mutation.SetProfitSafetyBuffer(v)
+	return _u
+}
+
+// SetNillableProfitSafetyBuffer sets the "profit_safety_buffer" field if the given value is not nil.
+func (_u *GroupUpdateOne) SetNillableProfitSafetyBuffer(v *float64) *GroupUpdateOne {
+	if v != nil {
+		_u.SetProfitSafetyBuffer(*v)
+	}
+	return _u
+}
+
+// AddProfitSafetyBuffer adds value to the "profit_safety_buffer" field.
+func (_u *GroupUpdateOne) AddProfitSafetyBuffer(v float64) *GroupUpdateOne {
+	_u.mutation.AddProfitSafetyBuffer(v)
+	return _u
+}
+
 // AddAPIKeyIDs adds the "api_keys" edge to the APIKey entity by IDs.
 func (_u *GroupUpdateOne) AddAPIKeyIDs(ids ...int64) *GroupUpdateOne {
 	_u.mutation.AddAPIKeyIDs(ids...)
@@ -3330,6 +3488,9 @@ func (_u *GroupUpdateOne) sqlSave(ctx context.Context) (_node *Group, err error)
 	if value, ok := _u.mutation.AllowMessagesDispatch(); ok {
 		_spec.SetField(group.FieldAllowMessagesDispatch, field.TypeBool, value)
 	}
+	if value, ok := _u.mutation.AllowLive(); ok {
+		_spec.SetField(group.FieldAllowLive, field.TypeBool, value)
+	}
 	if value, ok := _u.mutation.RequireOauthOnly(); ok {
 		_spec.SetField(group.FieldRequireOauthOnly, field.TypeBool, value)
 	}
@@ -3362,6 +3523,21 @@ func (_u *GroupUpdateOne) sqlSave(ctx context.Context) (_node *Group, err error)
 			sqljson.Append(u, group.FieldReasoningEffortMappings, value)
 		})
 	}
+	if value, ok := _u.mutation.ProfitControlEnabled(); ok {
+		_spec.SetField(group.FieldProfitControlEnabled, field.TypeBool, value)
+	}
+	if value, ok := _u.mutation.ProfitMinMargin(); ok {
+		_spec.SetField(group.FieldProfitMinMargin, field.TypeFloat64, value)
+	}
+	if value, ok := _u.mutation.AddedProfitMinMargin(); ok {
+		_spec.AddField(group.FieldProfitMinMargin, field.TypeFloat64, value)
+	}
+	if value, ok := _u.mutation.ProfitSafetyBuffer(); ok {
+		_spec.SetField(group.FieldProfitSafetyBuffer, field.TypeFloat64, value)
+	}
+	if value, ok := _u.mutation.AddedProfitSafetyBuffer(); ok {
+		_spec.AddField(group.FieldProfitSafetyBuffer, field.TypeFloat64, value)
+	}
 	if _u.mutation.APIKeysCleared() {
 		edge := &sqlgraph.EdgeSpec{
 			Rel:     sqlgraph.O2M,

backend/ent/migrate/schema.go

@@ -938,6 +938,7 @@ var (
 		{Name: "supported_model_scopes", Type: field.TypeJSON, SchemaType: map[string]string{"postgres": "jsonb"}},
 		{Name: "sort_order", Type: field.TypeInt, Default: 0},
 		{Name: "allow_messages_dispatch", Type: field.TypeBool, Default: false},
+		{Name: "allow_live", Type: field.TypeBool, Default: false},
 		{Name: "require_oauth_only", Type: field.TypeBool, Default: false},
 		{Name: "require_privacy_set", Type: field.TypeBool, Default: false},
 		{Name: "default_mapped_model", Type: field.TypeString, Size: 100, Default: ""},
@@ -946,6 +947,9 @@ var (
 		{Name: "rpm_limit", Type: field.TypeInt, Default: 0},
 		{Name: "max_reasoning_effort", Type: field.TypeString, Size: 20, Default: ""},
 		{Name: "reasoning_effort_mappings", Type: field.TypeJSON, SchemaType: map[string]string{"postgres": "jsonb"}},
+		{Name: "profit_control_enabled", Type: field.TypeBool, Default: false},
+		{Name: "profit_min_margin", Type: field.TypeFloat64, Default: 0, SchemaType: map[string]string{"postgres": "decimal(10,4)"}},
+		{Name: "profit_safety_buffer", Type: field.TypeFloat64, Default: 0, SchemaType: map[string]string{"postgres": "decimal(10,4)"}},
 	}
 	// GroupsTable holds the schema information for the "groups" table.
 	GroupsTable = &schema.Table{

backend/ent/mutation.go

@@ -21911,6 +21911,7 @@ type GroupMutation struct {
 	sort_order                              *int
 	addsort_order                           *int
 	allow_messages_dispatch                 *bool
+	allow_live                              *bool
 	require_oauth_only                      *bool
 	require_privacy_set                     *bool
 	default_mapped_model                    *string
@@ -21921,6 +21922,11 @@ type GroupMutation struct {
 	max_reasoning_effort                    *string
 	reasoning_effort_mappings               *[]domain.ReasoningEffortMapping
 	appendreasoning_effort_mappings         []domain.ReasoningEffortMapping
+	profit_control_enabled                  *bool
+	profit_min_margin                       *float64
+	addprofit_min_margin                    *float64
+	profit_safety_buffer                    *float64
+	addprofit_safety_buffer                 *float64
 	clearedFields                           map[string]struct{}
 	api_keys                                map[int64]struct{}
 	removedapi_keys                         map[int64]struct{}
@@ -24226,6 +24232,42 @@ func (m *GroupMutation) ResetAllowMessagesDispatch() {
 	m.allow_messages_dispatch = nil
 }
 
+// SetAllowLive sets the "allow_live" field.
+func (m *GroupMutation) SetAllowLive(b bool) {
+	m.allow_live = &b
+}
+
+// AllowLive returns the value of the "allow_live" field in the mutation.
+func (m *GroupMutation) AllowLive() (r bool, exists bool) {
+	v := m.allow_live
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// OldAllowLive returns the old "allow_live" field's value of the Group entity.
+// If the Group object wasn't provided to the builder, the object is fetched from the database.
+// An error is returned if the mutation operation is not UpdateOne, or the database query fails.
+func (m *GroupMutation) OldAllowLive(ctx context.Context) (v bool, err error) {
+	if !m.op.Is(OpUpdateOne) {
+		return v, errors.New("OldAllowLive is only allowed on UpdateOne operations")
+	}
+	if m.id == nil || m.oldValue == nil {
+		return v, errors.New("OldAllowLive requires an ID field in the mutation")
+	}
+	oldValue, err := m.oldValue(ctx)
+	if err != nil {
+		return v, fmt.Errorf("querying old value for OldAllowLive: %w", err)
+	}
+	return oldValue.AllowLive, nil
+}
+
+// ResetAllowLive resets all changes to the "allow_live" field.
+func (m *GroupMutation) ResetAllowLive() {
+	m.allow_live = nil
+}
+
 // SetRequireOauthOnly sets the "require_oauth_only" field.
 func (m *GroupMutation) SetRequireOauthOnly(b bool) {
 	m.require_oauth_only = &b
@@ -24549,6 +24591,154 @@ func (m *GroupMutation) ResetReasoningEffortMappings() {
 	m.appendreasoning_effort_mappings = nil
 }
 
+// SetProfitControlEnabled sets the "profit_control_enabled" field.
+func (m *GroupMutation) SetProfitControlEnabled(b bool) {
+	m.profit_control_enabled = &b
+}
+
+// ProfitControlEnabled returns the value of the "profit_control_enabled" field in the mutation.
+func (m *GroupMutation) ProfitControlEnabled() (r bool, exists bool) {
+	v := m.profit_control_enabled
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// OldProfitControlEnabled returns the old "profit_control_enabled" field's value of the Group entity.
+// If the Group object wasn't provided to the builder, the object is fetched from the database.
+// An error is returned if the mutation operation is not UpdateOne, or the database query fails.
+func (m *GroupMutation) OldProfitControlEnabled(ctx context.Context) (v bool, err error) {
+	if !m.op.Is(OpUpdateOne) {
+		return v, errors.New("OldProfitControlEnabled is only allowed on UpdateOne operations")
+	}
+	if m.id == nil || m.oldValue == nil {
+		return v, errors.New("OldProfitControlEnabled requires an ID field in the mutation")
+	}
+	oldValue, err := m.oldValue(ctx)
+	if err != nil {
+		return v, fmt.Errorf("querying old value for OldProfitControlEnabled: %w", err)
+	}
+	return oldValue.ProfitControlEnabled, nil
+}
+
+// ResetProfitControlEnabled resets all changes to the "profit_control_enabled" field.
+func (m *GroupMutation) ResetProfitControlEnabled() {
+	m.profit_control_enabled = nil
+}
+
+// SetProfitMinMargin sets the "profit_min_margin" field.
+func (m *GroupMutation) SetProfitMinMargin(f float64) {
+	m.profit_min_margin = &f
+	m.addprofit_min_margin = nil
+}
+
+// ProfitMinMargin returns the value of the "profit_min_margin" field in the mutation.
+func (m *GroupMutation) ProfitMinMargin() (r float64, exists bool) {
+	v := m.profit_min_margin
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// OldProfitMinMargin returns the old "profit_min_margin" field's value of the Group entity.
+// If the Group object wasn't provided to the builder, the object is fetched from the database.
+// An error is returned if the mutation operation is not UpdateOne, or the database query fails.
+func (m *GroupMutation) OldProfitMinMargin(ctx context.Context) (v float64, err error) {
+	if !m.op.Is(OpUpdateOne) {
+		return v, errors.New("OldProfitMinMargin is only allowed on UpdateOne operations")
+	}
+	if m.id == nil || m.oldValue == nil {
+		return v, errors.New("OldProfitMinMargin requires an ID field in the mutation")
+	}
+	oldValue, err := m.oldValue(ctx)
+	if err != nil {
+		return v, fmt.Errorf("querying old value for OldProfitMinMargin: %w", err)
+	}
+	return oldValue.ProfitMinMargin, nil
+}
+
+// AddProfitMinMargin adds f to the "profit_min_margin" field.
+func (m *GroupMutation) AddProfitMinMargin(f float64) {
+	if m.addprofit_min_margin != nil {
+		*m.addprofit_min_margin += f
+	} else {
+		m.addprofit_min_margin = &f
+	}
+}
+
+// AddedProfitMinMargin returns the value that was added to the "profit_min_margin" field in this mutation.
+func (m *GroupMutation) AddedProfitMinMargin() (r float64, exists bool) {
+	v := m.addprofit_min_margin
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// ResetProfitMinMargin resets all changes to the "profit_min_margin" field.
+func (m *GroupMutation) ResetProfitMinMargin() {
+	m.profit_min_margin = nil
+	m.addprofit_min_margin = nil
+}
+
+// SetProfitSafetyBuffer sets the "profit_safety_buffer" field.
+func (m *GroupMutation) SetProfitSafetyBuffer(f float64) {
+	m.profit_safety_buffer = &f
+	m.addprofit_safety_buffer = nil
+}
+
+// ProfitSafetyBuffer returns the value of the "profit_safety_buffer" field in the mutation.
+func (m *GroupMutation) ProfitSafetyBuffer() (r float64, exists bool) {
+	v := m.profit_safety_buffer
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// OldProfitSafetyBuffer returns the old "profit_safety_buffer" field's value of the Group entity.
+// If the Group object wasn't provided to the builder, the object is fetched from the database.
+// An error is returned if the mutation operation is not UpdateOne, or the database query fails.
+func (m *GroupMutation) OldProfitSafetyBuffer(ctx context.Context) (v float64, err error) {
+	if !m.op.Is(OpUpdateOne) {
+		return v, errors.New("OldProfitSafetyBuffer is only allowed on UpdateOne operations")
+	}
+	if m.id == nil || m.oldValue == nil {
+		return v, errors.New("OldProfitSafetyBuffer requires an ID field in the mutation")
+	}
+	oldValue, err := m.oldValue(ctx)
+	if err != nil {
+		return v, fmt.Errorf("querying old value for OldProfitSafetyBuffer: %w", err)
+	}
+	return oldValue.ProfitSafetyBuffer, nil
+}
+
+// AddProfitSafetyBuffer adds f to the "profit_safety_buffer" field.
+func (m *GroupMutation) AddProfitSafetyBuffer(f float64) {
+	if m.addprofit_safety_buffer != nil {
+		*m.addprofit_safety_buffer += f
+	} else {
+		m.addprofit_safety_buffer = &f
+	}
+}
+
+// AddedProfitSafetyBuffer returns the value that was added to the "profit_safety_buffer" field in this mutation.
+func (m *GroupMutation) AddedProfitSafetyBuffer() (r float64, exists bool) {
+	v := m.addprofit_safety_buffer
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// ResetProfitSafetyBuffer resets all changes to the "profit_safety_buffer" field.
+func (m *GroupMutation) ResetProfitSafetyBuffer() {
+	m.profit_safety_buffer = nil
+	m.addprofit_safety_buffer = nil
+}
+
 // AddAPIKeyIDs adds the "api_keys" edge to the APIKey entity by ids.
 func (m *GroupMutation) AddAPIKeyIDs(ids ...int64) {
 	if m.api_keys == nil {
@@ -24907,7 +25097,7 @@ func (m *GroupMutation) Type() string {
 // order to get all numeric fields that were incremented/decremented, call
 // AddedFields().
 func (m *GroupMutation) Fields() []string {
-	fields := make([]string, 0, 51)
+	fields := make([]string, 0, 55)
 	if m.created_at != nil {
 		fields = append(fields, group.FieldCreatedAt)
 	}
@@ -25037,6 +25227,9 @@ func (m *GroupMutation) Fields() []string {
 	if m.allow_messages_dispatch != nil {
 		fields = append(fields, group.FieldAllowMessagesDispatch)
 	}
+	if m.allow_live != nil {
+		fields = append(fields, group.FieldAllowLive)
+	}
 	if m.require_oauth_only != nil {
 		fields = append(fields, group.FieldRequireOauthOnly)
 	}
@@ -25061,6 +25254,15 @@ func (m *GroupMutation) Fields() []string {
 	if m.reasoning_effort_mappings != nil {
 		fields = append(fields, group.FieldReasoningEffortMappings)
 	}
+	if m.profit_control_enabled != nil {
+		fields = append(fields, group.FieldProfitControlEnabled)
+	}
+	if m.profit_min_margin != nil {
+		fields = append(fields, group.FieldProfitMinMargin)
+	}
+	if m.profit_safety_buffer != nil {
+		fields = append(fields, group.FieldProfitSafetyBuffer)
+	}
 	return fields
 }
 
@@ -25155,6 +25357,8 @@ func (m *GroupMutation) Field(name string) (ent.Value, bool) {
 		return m.SortOrder()
 	case group.FieldAllowMessagesDispatch:
 		return m.AllowMessagesDispatch()
+	case group.FieldAllowLive:
+		return m.AllowLive()
 	case group.FieldRequireOauthOnly:
 		return m.RequireOauthOnly()
 	case group.FieldRequirePrivacySet:
@@ -25171,6 +25375,12 @@ func (m *GroupMutation) Field(name string) (ent.Value, bool) {
 		return m.MaxReasoningEffort()
 	case group.FieldReasoningEffortMappings:
 		return m.ReasoningEffortMappings()
+	case group.FieldProfitControlEnabled:
+		return m.ProfitControlEnabled()
+	case group.FieldProfitMinMargin:
+		return m.ProfitMinMargin()
+	case group.FieldProfitSafetyBuffer:
+		return m.ProfitSafetyBuffer()
 	}
 	return nil, false
 }
@@ -25266,6 +25476,8 @@ func (m *GroupMutation) OldField(ctx context.Context, name string) (ent.Value, e
 		return m.OldSortOrder(ctx)
 	case group.FieldAllowMessagesDispatch:
 		return m.OldAllowMessagesDispatch(ctx)
+	case group.FieldAllowLive:
+		return m.OldAllowLive(ctx)
 	case group.FieldRequireOauthOnly:
 		return m.OldRequireOauthOnly(ctx)
 	case group.FieldRequirePrivacySet:
@@ -25282,6 +25494,12 @@ func (m *GroupMutation) OldField(ctx context.Context, name string) (ent.Value, e
 		return m.OldMaxReasoningEffort(ctx)
 	case group.FieldReasoningEffortMappings:
 		return m.OldReasoningEffortMappings(ctx)
+	case group.FieldProfitControlEnabled:
+		return m.OldProfitControlEnabled(ctx)
+	case group.FieldProfitMinMargin:
+		return m.OldProfitMinMargin(ctx)
+	case group.FieldProfitSafetyBuffer:
+		return m.OldProfitSafetyBuffer(ctx)
 	}
 	return nil, fmt.Errorf("unknown Group field %s", name)
 }
@@ -25592,6 +25810,13 @@ func (m *GroupMutation) SetField(name string, value ent.Value) error {
 		}
 		m.SetAllowMessagesDispatch(v)
 		return nil
+	case group.FieldAllowLive:
+		v, ok := value.(bool)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.SetAllowLive(v)
+		return nil
 	case group.FieldRequireOauthOnly:
 		v, ok := value.(bool)
 		if !ok {
@@ -25648,6 +25873,27 @@ func (m *GroupMutation) SetField(name string, value ent.Value) error {
 		}
 		m.SetReasoningEffortMappings(v)
 		return nil
+	case group.FieldProfitControlEnabled:
+		v, ok := value.(bool)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.SetProfitControlEnabled(v)
+		return nil
+	case group.FieldProfitMinMargin:
+		v, ok := value.(float64)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.SetProfitMinMargin(v)
+		return nil
+	case group.FieldProfitSafetyBuffer:
+		v, ok := value.(float64)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.SetProfitSafetyBuffer(v)
+		return nil
 	}
 	return fmt.Errorf("unknown Group field %s", name)
 }
@@ -25719,6 +25965,12 @@ func (m *GroupMutation) AddedFields() []string {
 	if m.addrpm_limit != nil {
 		fields = append(fields, group.FieldRpmLimit)
 	}
+	if m.addprofit_min_margin != nil {
+		fields = append(fields, group.FieldProfitMinMargin)
+	}
+	if m.addprofit_safety_buffer != nil {
+		fields = append(fields, group.FieldProfitSafetyBuffer)
+	}
 	return fields
 }
 
@@ -25769,6 +26021,10 @@ func (m *GroupMutation) AddedField(name string) (ent.Value, bool) {
 		return m.AddedSortOrder()
 	case group.FieldRpmLimit:
 		return m.AddedRpmLimit()
+	case group.FieldProfitMinMargin:
+		return m.AddedProfitMinMargin()
+	case group.FieldProfitSafetyBuffer:
+		return m.AddedProfitSafetyBuffer()
 	}
 	return nil, false
 }
@@ -25925,6 +26181,20 @@ func (m *GroupMutation) AddField(name string, value ent.Value) error {
 		}
 		m.AddRpmLimit(v)
 		return nil
+	case group.FieldProfitMinMargin:
+		v, ok := value.(float64)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.AddProfitMinMargin(v)
+		return nil
+	case group.FieldProfitSafetyBuffer:
+		v, ok := value.(float64)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.AddProfitSafetyBuffer(v)
+		return nil
 	}
 	return fmt.Errorf("unknown Group numeric field %s", name)
 }
@@ -26180,6 +26450,9 @@ func (m *GroupMutation) ResetField(name string) error {
 	case group.FieldAllowMessagesDispatch:
 		m.ResetAllowMessagesDispatch()
 		return nil
+	case group.FieldAllowLive:
+		m.ResetAllowLive()
+		return nil
 	case group.FieldRequireOauthOnly:
 		m.ResetRequireOauthOnly()
 		return nil
@@ -26204,6 +26477,15 @@ func (m *GroupMutation) ResetField(name string) error {
 	case group.FieldReasoningEffortMappings:
 		m.ResetReasoningEffortMappings()
 		return nil
+	case group.FieldProfitControlEnabled:
+		m.ResetProfitControlEnabled()
+		return nil
+	case group.FieldProfitMinMargin:
+		m.ResetProfitMinMargin()
+		return nil
+	case group.FieldProfitSafetyBuffer:
+		m.ResetProfitSafetyBuffer()
+		return nil
 	}
 	return fmt.Errorf("unknown Group field %s", name)
 }

backend/ent/runtime/runtime.go

@@ -1141,42 +1141,58 @@ func init() {
 	groupDescAllowMessagesDispatch := groupFields[39].Descriptor()
 	// group.DefaultAllowMessagesDispatch holds the default value on creation for the allow_messages_dispatch field.
 	group.DefaultAllowMessagesDispatch = groupDescAllowMessagesDispatch.Default.(bool)
+	// groupDescAllowLive is the schema descriptor for allow_live field.
+	groupDescAllowLive := groupFields[40].Descriptor()
+	// group.DefaultAllowLive holds the default value on creation for the allow_live field.
+	group.DefaultAllowLive = groupDescAllowLive.Default.(bool)
 	// groupDescRequireOauthOnly is the schema descriptor for require_oauth_only field.
-	groupDescRequireOauthOnly := groupFields[40].Descriptor()
+	groupDescRequireOauthOnly := groupFields[41].Descriptor()
 	// group.DefaultRequireOauthOnly holds the default value on creation for the require_oauth_only field.
 	group.DefaultRequireOauthOnly = groupDescRequireOauthOnly.Default.(bool)
 	// groupDescRequirePrivacySet is the schema descriptor for require_privacy_set field.
-	groupDescRequirePrivacySet := groupFields[41].Descriptor()
+	groupDescRequirePrivacySet := groupFields[42].Descriptor()
 	// group.DefaultRequirePrivacySet holds the default value on creation for the require_privacy_set field.
 	group.DefaultRequirePrivacySet = groupDescRequirePrivacySet.Default.(bool)
 	// groupDescDefaultMappedModel is the schema descriptor for default_mapped_model field.
-	groupDescDefaultMappedModel := groupFields[42].Descriptor()
+	groupDescDefaultMappedModel := groupFields[43].Descriptor()
 	// group.DefaultDefaultMappedModel holds the default value on creation for the default_mapped_model field.
 	group.DefaultDefaultMappedModel = groupDescDefaultMappedModel.Default.(string)
 	// group.DefaultMappedModelValidator is a validator for the "default_mapped_model" field. It is called by the builders before save.
 	group.DefaultMappedModelValidator = groupDescDefaultMappedModel.Validators[0].(func(string) error)
 	// groupDescMessagesDispatchModelConfig is the schema descriptor for messages_dispatch_model_config field.
-	groupDescMessagesDispatchModelConfig := groupFields[43].Descriptor()
+	groupDescMessagesDispatchModelConfig := groupFields[44].Descriptor()
 	// group.DefaultMessagesDispatchModelConfig holds the default value on creation for the messages_dispatch_model_config field.
 	group.DefaultMessagesDispatchModelConfig = groupDescMessagesDispatchModelConfig.Default.(domain.OpenAIMessagesDispatchModelConfig)
 	// groupDescModelsListConfig is the schema descriptor for models_list_config field.
-	groupDescModelsListConfig := groupFields[44].Descriptor()
+	groupDescModelsListConfig := groupFields[45].Descriptor()
 	// group.DefaultModelsListConfig holds the default value on creation for the models_list_config field.
 	group.DefaultModelsListConfig = groupDescModelsListConfig.Default.(domain.GroupModelsListConfig)
 	// groupDescRpmLimit is the schema descriptor for rpm_limit field.
-	groupDescRpmLimit := groupFields[45].Descriptor()
+	groupDescRpmLimit := groupFields[46].Descriptor()
 	// group.DefaultRpmLimit holds the default value on creation for the rpm_limit field.
 	group.DefaultRpmLimit = groupDescRpmLimit.Default.(int)
 	// groupDescMaxReasoningEffort is the schema descriptor for max_reasoning_effort field.
-	groupDescMaxReasoningEffort := groupFields[46].Descriptor()
+	groupDescMaxReasoningEffort := groupFields[47].Descriptor()
 	// group.DefaultMaxReasoningEffort holds the default value on creation for the max_reasoning_effort field.
 	group.DefaultMaxReasoningEffort = groupDescMaxReasoningEffort.Default.(string)
 	// group.MaxReasoningEffortValidator is a validator for the "max_reasoning_effort" field. It is called by the builders before save.
 	group.MaxReasoningEffortValidator = groupDescMaxReasoningEffort.Validators[0].(func(string) error)
 	// groupDescReasoningEffortMappings is the schema descriptor for reasoning_effort_mappings field.
-	groupDescReasoningEffortMappings := groupFields[47].Descriptor()
+	groupDescReasoningEffortMappings := groupFields[48].Descriptor()
 	// group.DefaultReasoningEffortMappings holds the default value on creation for the reasoning_effort_mappings field.
 	group.DefaultReasoningEffortMappings = groupDescReasoningEffortMappings.Default.([]domain.ReasoningEffortMapping)
+	// groupDescProfitControlEnabled is the schema descriptor for profit_control_enabled field.
+	groupDescProfitControlEnabled := groupFields[49].Descriptor()
+	// group.DefaultProfitControlEnabled holds the default value on creation for the profit_control_enabled field.
+	group.DefaultProfitControlEnabled = groupDescProfitControlEnabled.Default.(bool)
+	// groupDescProfitMinMargin is the schema descriptor for profit_min_margin field.
+	groupDescProfitMinMargin := groupFields[50].Descriptor()
+	// group.DefaultProfitMinMargin holds the default value on creation for the profit_min_margin field.
+	group.DefaultProfitMinMargin = groupDescProfitMinMargin.Default.(float64)
+	// groupDescProfitSafetyBuffer is the schema descriptor for profit_safety_buffer field.
+	groupDescProfitSafetyBuffer := groupFields[51].Descriptor()
+	// group.DefaultProfitSafetyBuffer holds the default value on creation for the profit_safety_buffer field.
+	group.DefaultProfitSafetyBuffer = groupDescProfitSafetyBuffer.Default.(float64)
 	idempotencyrecordMixin := schema.IdempotencyRecord{}.Mixin()
 	idempotencyrecordMixinFields0 := idempotencyrecordMixin[0].Fields()
 	_ = idempotencyrecordMixinFields0

backend/ent/schema/group.go

@@ -198,6 +198,9 @@ func (Group) Fields() []ent.Field {
 		field.Bool("allow_messages_dispatch").
 			Default(false).
 			Comment("是否允许 /v1/messages 调度到此 OpenAI 分组"),
+		field.Bool("allow_live").
+			Default(false).
+			Comment("是否允许此 OpenAI 分组访问 Live 接口"),
 		field.Bool("require_oauth_only").
 			Default(false).
 			Comment("仅允许非 apikey 类型账号关联到此分组"),
@@ -231,6 +234,20 @@ func (Group) Fields() []ent.Field {
 			Default([]domain.ReasoningEffortMapping{}).
 			SchemaType(map[string]string{dialect.Postgres: "jsonb"}).
 			Comment("OpenAI reasoning effort 自定义精确映射；先映射再应用上限"),
+
+		// 分组利润控制（migration 192/193）：openai/anthropic/gemini/grok/antigravity
+		// 的 token 分组可启用，composite 分组不能直接启用。
+		field.Bool("profit_control_enabled").
+			Default(false).
+			Comment("是否启用利润控制：调度时仅允许账号计费倍率满足毛利率要求的账号进入候选池"),
+		field.Float("profit_min_margin").
+			SchemaType(map[string]string{dialect.Postgres: "decimal(10,4)"}).
+			Default(0).
+			Comment("最低毛利率，小数（0.30=30%）；账号准入条件为 U <= D*(1-margin-buffer)"),
+		field.Float("profit_safety_buffer").
+			SchemaType(map[string]string{dialect.Postgres: "decimal(10,4)"}).
+			Default(0).
+			Comment("安全缓冲，小数；与 margin 相加后从下游倍率中扣除，默认 0"),
 	}
 }
 

backend/go.mod

@@ -16,11 +16,12 @@ require (
 	github.com/coder/websocket v1.8.14
 	github.com/dgraph-io/ristretto v0.2.0
 	github.com/gin-gonic/gin v1.9.1
-	github.com/golang-jwt/jwt/v5 v5.2.2
+	github.com/go-webauthn/webauthn v0.17.4
+	github.com/golang-jwt/jwt/v5 v5.3.1
 	github.com/google/uuid v1.6.0
 	github.com/google/wire v0.7.0
 	github.com/gorilla/websocket v1.5.3
-	github.com/imroc/req/v3 v3.57.0
+	github.com/imroc/req/v3 v3.59.0
 	github.com/klauspost/compress v1.18.2
 	github.com/lib/pq v1.10.9
 	github.com/patrickmn/go-cache v2.1.0+incompatible
@@ -43,7 +44,7 @@ require (
 	github.com/zeromicro/go-zero v1.9.4
 	go.uber.org/zap v1.24.0
 	golang.org/x/crypto v0.53.0
-	golang.org/x/image v0.39.0
+	golang.org/x/image v0.41.0
 	golang.org/x/mod v0.37.0
 	golang.org/x/net v0.56.0
 	golang.org/x/sync v0.21.0
@@ -97,6 +98,7 @@ require (
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/felixge/httpsnoop v1.0.4 // indirect
 	github.com/fsnotify/fsnotify v1.7.0 // indirect
+	github.com/fxamacker/cbor/v2 v2.9.2 // indirect
 	github.com/gabriel-vasile/mimetype v1.4.2 // indirect
 	github.com/gin-contrib/sse v0.1.0 // indirect
 	github.com/go-logr/logr v1.4.3 // indirect
@@ -106,9 +108,12 @@ require (
 	github.com/go-playground/locales v0.14.1 // indirect
 	github.com/go-playground/universal-translator v0.18.1 // indirect
 	github.com/go-playground/validator/v10 v10.14.0 // indirect
+	github.com/go-viper/mapstructure/v2 v2.5.0 // indirect
+	github.com/go-webauthn/x v0.2.6 // indirect
 	github.com/goccy/go-json v0.10.2 // indirect
 	github.com/google/go-cmp v0.7.0 // indirect
 	github.com/google/go-querystring v1.1.0 // indirect
+	github.com/google/go-tpm v0.9.8 // indirect
 	github.com/grpc-ecosystem/grpc-gateway/v2 v2.27.3 // indirect
 	github.com/hashicorp/hcl v1.0.0 // indirect
 	github.com/hashicorp/hcl/v2 v2.18.1 // indirect
@@ -137,11 +142,12 @@ require (
 	github.com/opencontainers/go-digest v1.0.0 // indirect
 	github.com/opencontainers/image-spec v1.1.1 // indirect
 	github.com/pelletier/go-toml/v2 v2.2.2 // indirect
+	github.com/philhofer/fwd v1.2.0 // indirect
 	github.com/pkg/errors v0.9.1 // indirect
 	github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect
 	github.com/power-devops/perfstat v0.0.0-20210106213030-5aafc221ea8c // indirect
 	github.com/quic-go/qpack v0.6.0 // indirect
-	github.com/quic-go/quic-go v0.57.1 // indirect
+	github.com/quic-go/quic-go v0.60.0 // indirect
 	github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect
 	github.com/sagikazarmark/locafero v0.4.0 // indirect
 	github.com/sagikazarmark/slog-shim v0.1.0 // indirect
@@ -158,20 +164,22 @@ require (
 	github.com/testcontainers/testcontainers-go v0.40.0 // indirect
 	github.com/tidwall/match v1.1.1 // indirect
 	github.com/tidwall/pretty v1.2.0 // indirect
+	github.com/tinylib/msgp v1.6.4 // indirect
 	github.com/tklauser/go-sysconf v0.3.12 // indirect
 	github.com/tklauser/numcpus v0.6.1 // indirect
 	github.com/twitchyliquid64/golang-asm v0.15.1 // indirect
 	github.com/ugorji/go/codec v1.2.11 // indirect
+	github.com/x448/float16 v0.8.4 // indirect
 	github.com/yuin/gopher-lua v1.1.1 // indirect
 	github.com/yusufpapurcu/wmi v1.2.4 // indirect
 	github.com/zclconf/go-cty v1.14.4 // indirect
 	github.com/zclconf/go-cty-yaml v1.1.0 // indirect
-	go.opentelemetry.io/auto/sdk v1.1.0 // indirect
+	go.opentelemetry.io/auto/sdk v1.2.1 // indirect
 	go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.49.0 // indirect
-	go.opentelemetry.io/otel v1.37.0 // indirect
-	go.opentelemetry.io/otel/metric v1.37.0 // indirect
-	go.opentelemetry.io/otel/sdk v1.37.0 // indirect
-	go.opentelemetry.io/otel/trace v1.37.0 // indirect
+	go.opentelemetry.io/otel v1.41.0 // indirect
+	go.opentelemetry.io/otel/metric v1.41.0 // indirect
+	go.opentelemetry.io/otel/sdk v1.41.0 // indirect
+	go.opentelemetry.io/otel/trace v1.41.0 // indirect
 	go.uber.org/atomic v1.10.0 // indirect
 	go.uber.org/automaxprocs v1.6.0 // indirect
 	go.uber.org/multierr v1.9.0 // indirect
@@ -179,6 +187,7 @@ require (
 	golang.org/x/exp v0.0.0-20251023183803-a4bb9ffd2546 // indirect
 	golang.org/x/sys v0.46.0 // indirect
 	golang.org/x/text v0.39.0 // indirect
+	golang.org/x/time v0.12.0 // indirect
 	golang.org/x/tools v0.47.0 // indirect
 	google.golang.org/grpc v1.75.1 // indirect
 	google.golang.org/protobuf v1.36.10 // indirect

backend/go.sum

@@ -128,6 +128,8 @@ github.com/frankban/quicktest v1.14.6 h1:7Xjx+VpznH+oBnejlPUj8oUpdxnVs4f8XU8WnHk
 github.com/frankban/quicktest v1.14.6/go.mod h1:4ptaffx2x8+WTWXmUCuVU6aPUX1/Mz7zb5vbUoiM6w0=
 github.com/fsnotify/fsnotify v1.7.0 h1:8JEhPFa5W2WU7YfeZzPNqzMP6Lwt7L2715Ggo0nosvA=
 github.com/fsnotify/fsnotify v1.7.0/go.mod h1:40Bi/Hjc2AVfZrqy+aj+yEI+/bRxZnMJyTJwOpGvigM=
+github.com/fxamacker/cbor/v2 v2.9.2 h1:X4Ksno9+x3cz0TZv69ec1hxP/+tymuR8PXQJyDwfh78=
+github.com/fxamacker/cbor/v2 v2.9.2/go.mod h1:vM4b+DJCtHn+zz7h3FFp/hDAI9WNWCsZj23V5ytsSxQ=
 github.com/gabriel-vasile/mimetype v1.4.2 h1:w5qFW6JKBz9Y393Y4q372O9A7cUSequkh1Q7OhCmWKU=
 github.com/gabriel-vasile/mimetype v1.4.2/go.mod h1:zApsH/mKG4w07erKIaJPFiX0Tsq9BFQgN3qGY5GnNgA=
 github.com/gin-contrib/sse v0.1.0 h1:Y/yl/+YNO8GZSjAhjMsSuLt29uWRFHdHYUb5lYOV9qE=
@@ -153,21 +155,29 @@ github.com/go-playground/validator/v10 v10.14.0 h1:vgvQWe3XCz3gIeFDm/HnTIbj6UGmg
 github.com/go-playground/validator/v10 v10.14.0/go.mod h1:9iXMNT7sEkjXb0I+enO7QXmzG6QCsPWY4zveKFVRSyU=
 github.com/go-test/deep v1.0.3 h1:ZrJSEWsXzPOxaZnFteGEfooLba+ju3FYIbOrS+rQd68=
 github.com/go-test/deep v1.0.3/go.mod h1:wGDj63lr65AM2AQyKZd/NYHGb0R+1RLqB8NKt3aSFNA=
+github.com/go-viper/mapstructure/v2 v2.5.0 h1:vM5IJoUAy3d7zRSVtIwQgBj7BiWtMPfmPEgAXnvj1Ro=
+github.com/go-viper/mapstructure/v2 v2.5.0/go.mod h1:oJDH3BJKyqBA2TXFhDsKDGDTlndYOZ6rGS0BRZIxGhM=
+github.com/go-webauthn/webauthn v0.17.4 h1:KFTSz3R2RYDiUn/0cDi3XTJgFenSG74eKTTHlqWhlxk=
+github.com/go-webauthn/webauthn v0.17.4/go.mod h1:pZk63EE/BdztlmyS4Yc+9H5g4a8blNlbtGmdHQHbZX8=
+github.com/go-webauthn/x v0.2.6 h1:TEyDuQAIiEgYpx60nKiBJIX/5nSUC8LxNbH+uf5U9uk=
+github.com/go-webauthn/x v0.2.6/go.mod h1:45bA7YEqyQhRcQJ/TiBb46Ww8yqHBGvgEhQ3WWF0aDo=
 github.com/goccy/go-json v0.10.2 h1:CrxCmQqYDkv1z7lO7Wbh2HN93uovUHgrECaO5ZrCXAU=
 github.com/goccy/go-json v0.10.2/go.mod h1:6MelG93GURQebXPDq3khkgXZkazVtN9CRI+MGFi0w8I=
 github.com/golang-jwt/jwt/v5 v5.2.2 h1:Rl4B7itRWVtYIHFrSNd7vhTiz9UpLdi6gZhZ3wEeDy8=
 github.com/golang-jwt/jwt/v5 v5.2.2/go.mod h1:pqrtFR0X4osieyHYxtmOUWsAWrfe1Q5UVIyoH402zdk=
+github.com/golang-jwt/jwt/v5 v5.3.1 h1:kYf81DTWFe7t+1VvL7eS+jKFVWaUnK9cB1qbwn63YCY=
+github.com/golang-jwt/jwt/v5 v5.3.1/go.mod h1:fxCRLWMO43lRc8nhHWY6LGqRcf+1gQWArsqaEUEa5bE=
 github.com/google/go-cmp v0.5.2/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.5.6/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
 github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
 github.com/google/go-querystring v1.1.0 h1:AnCroh3fv4ZBgVIf1Iwtovgjaw/GiKJo8M8yD/fhyJ8=
 github.com/google/go-querystring v1.1.0/go.mod h1:Kcdr2DB4koayq7X8pmAG4sNG59So17icRSOU623lUBU=
+github.com/google/go-tpm v0.9.8 h1:slArAR9Ft+1ybZu0lBwpSmpwhRXaa85hWtMinMyRAWo=
+github.com/google/go-tpm v0.9.8/go.mod h1:h9jEsEECg7gtLis0upRBQU+GhYVH6jMjrFxI8u6bVUY=
 github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/M65Eg=
 github.com/google/pprof v0.0.0-20250317173921-a4b03ec1a45e h1:ijClszYn+mADRFY17kjQEVQ1XRhq2/JR1M3sGqeJoxs=
 github.com/google/pprof v0.0.0-20250317173921-a4b03ec1a45e/go.mod h1:boTsfXsheKC2y+lKOCMpSfarhxDeIzfZG1jqGcPl3cA=
-github.com/google/subcommands v1.2.0 h1:vWQspBTo2nEqTUFita5/KeEWlUL8kQObDFbub/EN9oE=
-github.com/google/subcommands v1.2.0/go.mod h1:ZjhPrFU+Olkh9WazFPsl27BQ4UPiG37m3yTrtFlrHVk=
 github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
 github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
 github.com/google/wire v0.7.0 h1:JxUKI6+CVBgCO2WToKy/nQk0sS+amI9z9EjVmdaocj4=
@@ -185,8 +195,8 @@ github.com/hashicorp/hcl/v2 v2.18.1 h1:6nxnOJFku1EuSawSD81fuviYUV8DxFr3fp2dUi3ZY
 github.com/hashicorp/hcl/v2 v2.18.1/go.mod h1:ThLC89FV4p9MPW804KVbe/cEXoQ8NZEh+JtMeeGErHE=
 github.com/icholy/digest v1.1.0 h1:HfGg9Irj7i+IX1o1QAmPfIBNu/Q5A5Tu3n/MED9k9H4=
 github.com/icholy/digest v1.1.0/go.mod h1:QNrsSGQ5v7v9cReDI0+eyjsXGUoRSUZQHeQ5C4XLa0Y=
-github.com/imroc/req/v3 v3.57.0 h1:LMTUjNRUybUkTPn8oJDq8Kg3JRBOBTcnDhKu7mzupKI=
-github.com/imroc/req/v3 v3.57.0/go.mod h1:JL62ey1nvSLq81HORNcosvlf7SxZStONNqOprg0Pz00=
+github.com/imroc/req/v3 v3.59.0 h1:PqKhJHyBmJYob47LVuTHwRZE00ZO6icbLHe5Zra13jo=
+github.com/imroc/req/v3 v3.59.0/go.mod h1:cwLwwaE90Pq6ZMmKuHA9FivJbg3avnCH8/RA5lUUs74=
 github.com/jackc/pgpassfile v1.0.0 h1:/6Hmqy13Ss2zCq62VdNG8tM1wchn8zjSGOBJ6icpsIM=
 github.com/jackc/pgpassfile v1.0.0/go.mod h1:CEx0iS5ambNFdcRtxPj5JhEz+xB6uRky5eyVu/W2HEg=
 github.com/jackc/pgservicefile v0.0.0-20240606120523-5a60cdf6a761 h1:iCEnooe7UlwOQYpKFhBabPMi4aNAfoODPEFNiAnClxo=
@@ -222,8 +232,6 @@ github.com/mattn/go-colorable v0.1.13/go.mod h1:7S9/ev0klgBDR4GtXTXX8a3vIGJpMovk
 github.com/mattn/go-isatty v0.0.16/go.mod h1:kYGgaQfpe5nmfYZH+SKPsOc2e4SrIfOl2e/yFXSvRLM=
 github.com/mattn/go-isatty v0.0.20 h1:xfD0iDuEKnDkl03q4limB+vH+GxLEtL/jb4xVJSWWEY=
 github.com/mattn/go-isatty v0.0.20/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
-github.com/mattn/go-runewidth v0.0.15 h1:UNAjwbU9l54TA3KzvqLGxwWjHmMgBUVhBiTjelZgg3U=
-github.com/mattn/go-runewidth v0.0.15/go.mod h1:Jdepj2loyihRzMpdS35Xk/zdY8IAYHsh153qUoGf23w=
 github.com/mattn/go-sqlite3 v1.14.17 h1:mCRHCLDUBXgpKAqIKsaAaAsrAlbkeomtRFKXh2L6YIM=
 github.com/mattn/go-sqlite3 v1.14.17/go.mod h1:2eHXhiwb8IkHr+BDWZGa96P6+rkvnG63S2DGjv9HUNg=
 github.com/mdelapenya/tlscert v0.2.0 h1:7H81W6Z/4weDvZBNOfQte5GpIMo0lGYEeWbkGp5LJHI=
@@ -257,8 +265,6 @@ github.com/morikuni/aec v1.0.0 h1:nP9CBfwrvYnBRgY6qfDQkygYDmYwOilePFkwzv4dU8A=
 github.com/morikuni/aec v1.0.0/go.mod h1:BbKIizmSmc5MMPqRYbxO4ZU0S0+P200+tUnFx7PXmsc=
 github.com/ncruces/go-strftime v1.0.0 h1:HMFp8mLCTPp341M/ZnA4qaf7ZlsbTc+miZjCLOFAw7w=
 github.com/ncruces/go-strftime v1.0.0/go.mod h1:Fwc5htZGVVkseilnfgOVb9mKy6w1naJmn9CehxcKcls=
-github.com/olekukonko/tablewriter v0.0.5 h1:P2Ga83D34wi1o9J6Wh1mRuqd4mF/x/lgBS7N7AbDhec=
-github.com/olekukonko/tablewriter v0.0.5/go.mod h1:hPp6KlRPjbx+hW8ykQs1w3UBbZlj6HuIJcUGPhkA7kY=
 github.com/opencontainers/go-digest v1.0.0 h1:apOUWs51W5PlhuyGyz9FCeeBIOUDA/6nW8Oi/yOhh5U=
 github.com/opencontainers/go-digest v1.0.0/go.mod h1:0JzlMkj0TRzQZfJkVvzbP0HBR3IKzErnv2BNG4W4MAM=
 github.com/opencontainers/image-spec v1.1.1 h1:y0fUlFfIZhPF1W537XOLg0/fcx6zcHCJwooC2xJA040=
@@ -267,6 +273,8 @@ github.com/patrickmn/go-cache v2.1.0+incompatible h1:HRMgzkcYKYpi3C8ajMPV8OFXaaR
 github.com/patrickmn/go-cache v2.1.0+incompatible/go.mod h1:3Qf8kWWT7OJRJbdiICTKqZju1ZixQ/KpMGzzAfe6+WQ=
 github.com/pelletier/go-toml/v2 v2.2.2 h1:aYUidT7k73Pcl9nb2gScu7NSrKCSHIDE89b3+6Wq+LM=
 github.com/pelletier/go-toml/v2 v2.2.2/go.mod h1:1t835xjRzz80PqgE6HHgN2JOsmgYu/h4qDAS4n929Rs=
+github.com/philhofer/fwd v1.2.0 h1:e6DnBTl7vGY+Gz322/ASL4Gyp1FspeMvx1RNDoToZuM=
+github.com/philhofer/fwd v1.2.0/go.mod h1:RqIHx9QI14HlwKwm98g9Re5prTQ6LdeRQn+gXJFxsJM=
 github.com/pkg/errors v0.9.1 h1:FEBLx1zS214owpjy7qsBeixbURkuhQAwrK5UwLGTwt4=
 github.com/pkg/errors v0.9.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
@@ -278,22 +286,22 @@ github.com/pquerna/otp v1.5.0 h1:NMMR+WrmaqXU4EzdGJEE1aUUI0AMRzsp96fFFWNPwxs=
 github.com/pquerna/otp v1.5.0/go.mod h1:dkJfzwRKNiegxyNb54X/3fLwhCynbMspSyWKnvi1AEg=
 github.com/prashantv/gostub v1.1.0 h1:BTyx3RfQjRHnUWaGF9oQos79AlQ5k8WNktv7VGvVH4g=
 github.com/prashantv/gostub v1.1.0/go.mod h1:A5zLQHz7ieHGG7is6LLXLz7I8+3LZzsrV0P1IAHhP5U=
+github.com/quic-go/go-ossfuzz-seeds v0.1.0 h1:APacT+iIaNF6fd8AGEiN3bT/Jtkd2jz4v4TzM7MFjy0=
+github.com/quic-go/go-ossfuzz-seeds v0.1.0/go.mod h1:3IOHRbJIc+L6YKMwfDtJAM9Vj9k0YY4muhuyUYk5tbk=
 github.com/quic-go/qpack v0.6.0 h1:g7W+BMYynC1LbYLSqRt8PBg5Tgwxn214ZZR34VIOjz8=
 github.com/quic-go/qpack v0.6.0/go.mod h1:lUpLKChi8njB4ty2bFLX2x4gzDqXwUpaO1DP9qMDZII=
-github.com/quic-go/quic-go v0.57.1 h1:25KAAR9QR8KZrCZRThWMKVAwGoiHIrNbT72ULHTuI10=
-github.com/quic-go/quic-go v0.57.1/go.mod h1:ly4QBAjHA2VhdnxhojRsCUOeJwKYg+taDlos92xb1+s=
+github.com/quic-go/quic-go v0.60.0 h1:xcQioE8OM66UQLeUMHltK1CCcOu3JbVB4JAQdDQSB+0=
+github.com/quic-go/quic-go v0.60.0/go.mod h1:wpKpjmPpftl30sL6pFh7REVpjbcCVy4zt2vDyK1TuJk=
 github.com/redis/go-redis/v9 v9.17.2 h1:P2EGsA4qVIM3Pp+aPocCJ7DguDHhqrXNhVcEp4ViluI=
 github.com/redis/go-redis/v9 v9.17.2/go.mod h1:u410H11HMLoB+TP67dz8rL9s6QW2j76l0//kSOd3370=
 github.com/refraction-networking/utls v1.8.2 h1:j4Q1gJj0xngdeH+Ox/qND11aEfhpgoEvV+S9iJ2IdQo=
 github.com/refraction-networking/utls v1.8.2/go.mod h1:jkSOEkLqn+S/jtpEHPOsVv/4V4EVnelwbMQl4vCWXAM=
 github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec h1:W09IVJc94icq4NjY3clb7Lk8O1qJ8BdBEF8z0ibU0rE=
 github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec/go.mod h1:qqbHyh8v60DhA7CoWK5oRCqLrMHRGoxYCSS9EjAz6Eo=
-github.com/rivo/uniseg v0.2.0 h1:S1pD9weZBuJdFmowNwbpi7BJ8TNftyUImj/0WQi72jY=
-github.com/rivo/uniseg v0.2.0/go.mod h1:J6wj4VEh+S6ZtnVlnTBMWIodfgj8LQOQFoIToxlJtxc=
 github.com/robfig/cron/v3 v3.0.1 h1:WdRxkvbJztn8LMz/QEvLN5sBU+xKpSqwwUO1Pjr4qDs=
 github.com/robfig/cron/v3 v3.0.1/go.mod h1:eQICP3HwyT7UooqI/z+Ov+PtYAWygg1TEWWzGIFLtro=
-github.com/rogpeppe/go-internal v1.13.1 h1:KvO1DLK/DRN07sQ1LQKScxyZJuNnedQ5/wKSR38lUII=
-github.com/rogpeppe/go-internal v1.13.1/go.mod h1:uMEvuHeurkdAXX61udpOXGD/AzZDWNMNyH2VO9fmH0o=
+github.com/rogpeppe/go-internal v1.14.1 h1:UQB4HGPB6osV0SQTLymcB4TgvyWu6ZyliaW0tI/otEQ=
+github.com/rogpeppe/go-internal v1.14.1/go.mod h1:MaRKkUm5W0goXpeCfT7UZI6fk/L7L7so1lCWt35ZSgc=
 github.com/sagikazarmark/locafero v0.4.0 h1:HApY1R9zGo4DBgr7dqsTH/JJxLTTsOt7u6keLGt6kNQ=
 github.com/sagikazarmark/locafero v0.4.0/go.mod h1:Pe1W6UlPYUk/+wc/6KFhbORCfqzgYEpgQ3O5fPuL3H4=
 github.com/sagikazarmark/slog-shim v0.1.0 h1:diDBnUNK9N/354PgrxMywXnAwEr1QZcOr6gto+ugjYE=
@@ -322,8 +330,6 @@ github.com/spf13/afero v1.11.0 h1:WJQKhtpdm3v2IzqG8VMqrr6Rf3UYpEF239Jy9wNepM8=
 github.com/spf13/afero v1.11.0/go.mod h1:GH9Y3pIexgf1MTIWtNGyogA5MwRIDXGUr+hbWNoBjkY=
 github.com/spf13/cast v1.6.0 h1:GEiTHELF+vaR5dhz3VqZfFSzZjYbgeKDpBxQVS4GYJ0=
 github.com/spf13/cast v1.6.0/go.mod h1:ancEpBxwJDODSW/UG4rDrAqiKolqNNh2DX3mk86cAdo=
-github.com/spf13/cobra v1.7.0 h1:hyqWnYt1ZQShIddO5kBpj3vu05/++x6tJ6dg8EC572I=
-github.com/spf13/cobra v1.7.0/go.mod h1:uLxZILRyS/50WlhOIKD7W6V5bgeIt+4sICxh6uRMrb0=
 github.com/spf13/pflag v1.0.5 h1:iy+VFUOCP1a+8yFto/drg2CJ5u0yRoB7fZw3DKv/JXA=
 github.com/spf13/pflag v1.0.5/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
 github.com/spf13/viper v1.18.2 h1:LUXCnvUvSM6FXAsj6nnfc8Q2tp1dIgUfY9Kc8GsSOiQ=
@@ -364,6 +370,8 @@ github.com/tidwall/sjson v1.2.5 h1:kLy8mja+1c9jlljvWTlSazM7cKDRfJuR/bOJhcY5NcY=
 github.com/tidwall/sjson v1.2.5/go.mod h1:Fvgq9kS/6ociJEDnK0Fk1cpYF4FIW6ZF7LAe+6jwd28=
 github.com/tiktoken-go/tokenizer v0.8.0 h1:drHWno2Zx3eAm/hk/LmvBKXPpSImB7BRyh/ru4+3Q7Y=
 github.com/tiktoken-go/tokenizer v0.8.0/go.mod h1:pTmPz4r14MV3JkUGAmAcdLdYhSxN68MCjrP+EoxBdx0=
+github.com/tinylib/msgp v1.6.4 h1:mOwYbyYDLPj35mkA2BjjYejgJk9BuHxDdvRnb6v2ZcQ=
+github.com/tinylib/msgp v1.6.4/go.mod h1:RSp0LW9oSxFut3KzESt5Voq4GVWyS+PSulT77roAqEA=
 github.com/tklauser/go-sysconf v0.3.12 h1:0QaGUFOdQaIVdPgfITYzaTegZvdCjmYO52cSFAEVmqU=
 github.com/tklauser/go-sysconf v0.3.12/go.mod h1:Ho14jnntGE1fpdOqQEEaiKRpvIavV0hSfmBq8nJbHYI=
 github.com/tklauser/numcpus v0.6.1 h1:ng9scYS7az0Bk4OZLvrNXNSAO2Pxr1XXRAPyjhIx+Fk=
@@ -374,6 +382,8 @@ github.com/ugorji/go/codec v1.2.11 h1:BMaWp1Bb6fHwEtbplGBGJ498wD+LKlNSl25MjdZY4d
 github.com/ugorji/go/codec v1.2.11/go.mod h1:UNopzCgEMSXjBc6AOMqYvWC1ktqTAfzJZUZgYf6w6lg=
 github.com/wechatpay-apiv3/wechatpay-go v0.2.21 h1:uIyMpzvcaHA33W/QPtHstccw+X52HO1gFdvVL9O6Lfs=
 github.com/wechatpay-apiv3/wechatpay-go v0.2.21/go.mod h1:A254AUBVB6R+EqQFo3yTgeh7HtyqRRtN2w9hQSOrd4Q=
+github.com/x448/float16 v0.8.4 h1:qLwI1I70+NjRFUR3zs1JPUCgaCXSh3SW62uAKT1mSBM=
+github.com/x448/float16 v0.8.4/go.mod h1:14CWIYCyZA/cWjXOioeEpHeN/83MdbZDRQHoFcYsOfg=
 github.com/xyproto/randomstring v1.0.5 h1:YtlWPoRdgMu3NZtP45drfy1GKoojuR7hmRcnhZqKjWU=
 github.com/xyproto/randomstring v1.0.5/go.mod h1:rgmS5DeNXLivK7YprL0pY+lTuhNQW3iGxZ18UQApw/E=
 github.com/yuin/gopher-lua v1.1.1 h1:kYKnWBjvbNP4XLT3+bPEwAXJx262OhaHDWDVOPjL46M=
@@ -386,22 +396,22 @@ github.com/zclconf/go-cty-yaml v1.1.0 h1:nP+jp0qPHv2IhUVqmQSzjvqAWcObN0KBkUl2rWB
 github.com/zclconf/go-cty-yaml v1.1.0/go.mod h1:9YLUH4g7lOhVWqUbctnVlZ5KLpg7JAprQNgxSZ1Gyxs=
 github.com/zeromicro/go-zero v1.9.4 h1:aRLFoISqAYijABtkbliQC5SsI5TbizJpQvoHc9xup8k=
 github.com/zeromicro/go-zero v1.9.4/go.mod h1:a17JOTch25SWxBcUgJZYps60hygK3pIYdw7nGwlcS38=
-go.opentelemetry.io/auto/sdk v1.1.0 h1:cH53jehLUN6UFLY71z+NDOiNJqDdPRaXzTel0sJySYA=
-go.opentelemetry.io/auto/sdk v1.1.0/go.mod h1:3wSPjt5PWp2RhlCcmmOial7AvC4DQqZb7a7wCow3W8A=
+go.opentelemetry.io/auto/sdk v1.2.1 h1:jXsnJ4Lmnqd11kwkBV2LgLoFMZKizbCi5fNZ/ipaZ64=
+go.opentelemetry.io/auto/sdk v1.2.1/go.mod h1:KRTj+aOaElaLi+wW1kO/DZRXwkF4C5xPbEe3ZiIhN7Y=
 go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.49.0 h1:jq9TW8u3so/bN+JPT166wjOI6/vQPF6Xe7nMNIltagk=
 go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.49.0/go.mod h1:p8pYQP+m5XfbZm9fxtSKAbM6oIllS7s2AfxrChvc7iw=
-go.opentelemetry.io/otel v1.37.0 h1:9zhNfelUvx0KBfu/gb+ZgeAfAgtWrfHJZcAqFC228wQ=
-go.opentelemetry.io/otel v1.37.0/go.mod h1:ehE/umFRLnuLa/vSccNq9oS1ErUlkkK71gMcN34UG8I=
+go.opentelemetry.io/otel v1.41.0 h1:YlEwVsGAlCvczDILpUXpIpPSL/VPugt7zHThEMLce1c=
+go.opentelemetry.io/otel v1.41.0/go.mod h1:Yt4UwgEKeT05QbLwbyHXEwhnjxNO6D8L5PQP51/46dE=
 go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.24.0 h1:t6wl9SPayj+c7lEIFgm4ooDBZVb01IhLB4InpomhRw8=
 go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.24.0/go.mod h1:iSDOcsnSA5INXzZtwaBPrKp/lWu/V14Dd+llD0oI2EA=
 go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.24.0 h1:Xw8U6u2f8DK2XAkGRFV7BBLENgnTGX9i4rQRxJf+/vs=
 go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracehttp v1.24.0/go.mod h1:6KW1Fm6R/s6Z3PGXwSJN2K4eT6wQB3vXX6CVnYX9NmM=
-go.opentelemetry.io/otel/metric v1.37.0 h1:mvwbQS5m0tbmqML4NqK+e3aDiO02vsf/WgbsdpcPoZE=
-go.opentelemetry.io/otel/metric v1.37.0/go.mod h1:04wGrZurHYKOc+RKeye86GwKiTb9FKm1WHtO+4EVr2E=
-go.opentelemetry.io/otel/sdk v1.37.0 h1:ItB0QUqnjesGRvNcmAcU0LyvkVyGJ2xftD29bWdDvKI=
-go.opentelemetry.io/otel/sdk v1.37.0/go.mod h1:VredYzxUvuo2q3WRcDnKDjbdvmO0sCzOvVAiY+yUkAg=
-go.opentelemetry.io/otel/trace v1.37.0 h1:HLdcFNbRQBE2imdSEgm/kwqmQj1Or1l/7bW6mxVK7z4=
-go.opentelemetry.io/otel/trace v1.37.0/go.mod h1:TlgrlQ+PtQO5XFerSPUYG0JSgGyryXewPGyayAWSBS0=
+go.opentelemetry.io/otel/metric v1.41.0 h1:rFnDcs4gRzBcsO9tS8LCpgR0dxg4aaxWlJxCno7JlTQ=
+go.opentelemetry.io/otel/metric v1.41.0/go.mod h1:xPvCwd9pU0VN8tPZYzDZV/BMj9CM9vs00GuBjeKhJps=
+go.opentelemetry.io/otel/sdk v1.41.0 h1:YPIEXKmiAwkGl3Gu1huk1aYWwtpRLeskpV+wPisxBp8=
+go.opentelemetry.io/otel/sdk v1.41.0/go.mod h1:ahFdU0G5y8IxglBf0QBJXgSe7agzjE4GiTJ6HT9ud90=
+go.opentelemetry.io/otel/trace v1.41.0 h1:Vbk2co6bhj8L59ZJ6/xFTskY+tGAbOnCtQGVVa9TIN0=
+go.opentelemetry.io/otel/trace v1.41.0/go.mod h1:U1NU4ULCoxeDKc09yCWdWe+3QoyweJcISEVa1RBzOis=
 go.opentelemetry.io/proto/otlp v1.3.1 h1:TrMUixzpM0yuc/znrFTP9MMRh8trP93mkCiDVeXrui0=
 go.opentelemetry.io/proto/otlp v1.3.1/go.mod h1:0X1WI4de4ZsLrrJNLAQbFeLCm3T7yBkR0XqQ7niQU+8=
 go.uber.org/atomic v1.10.0 h1:9qC72Qh0+3MqyJbAn8YU5xVq1frD8bn3JtD2oXtafVQ=
@@ -419,24 +429,16 @@ go.uber.org/zap v1.24.0/go.mod h1:2kMP+WWQ8aoFoedH3T2sq6iJ2yDWpHbP0f6MQbS9Gkg=
 golang.org/x/arch v0.0.0-20210923205945-b76863e36670/go.mod h1:5om86z9Hs0C8fWVUuoMHwpExlXzs5Tkyp9hOrfG7pp8=
 golang.org/x/arch v0.3.0 h1:02VY4/ZcO/gBOH6PUaoiptASxtXU10jazRCP865E97k=
 golang.org/x/arch v0.3.0/go.mod h1:5om86z9Hs0C8fWVUuoMHwpExlXzs5Tkyp9hOrfG7pp8=
-golang.org/x/crypto v0.51.0 h1:IBPXwPfKxY7cWQZ38ZCIRPI50YLeevDLlLnyC5wRGTI=
-golang.org/x/crypto v0.51.0/go.mod h1:8AdwkbraGNABw2kOX6YFPs3WM22XqI4EXEd8g+x7Oc8=
 golang.org/x/crypto v0.53.0 h1:QZ4Muo8THX6CizN2vPPd5fBGHyogrdK9fG4wLPFUsto=
 golang.org/x/crypto v0.53.0/go.mod h1:DNLU434OwVakk9PzuwV8w62mAJpRJL3vsgcfp4Qnsio=
 golang.org/x/exp v0.0.0-20251023183803-a4bb9ffd2546 h1:mgKeJMpvi0yx/sU5GsxQ7p6s2wtOnGAHZWCHUM4KGzY=
 golang.org/x/exp v0.0.0-20251023183803-a4bb9ffd2546/go.mod h1:j/pmGrbnkbPtQfxEe5D0VQhZC6qKbfKifgD0oM7sR70=
-golang.org/x/image v0.39.0 h1:skVYidAEVKgn8lZ602XO75asgXBgLj9G/FE3RbuPFww=
-golang.org/x/image v0.39.0/go.mod h1:sIbmppfU+xFLPIG0FoVUTvyBMmgng1/XAMhQ2ft0hpA=
-golang.org/x/mod v0.35.0 h1:Ww1D637e6Pg+Zb2KrWfHQUnH2dQRLBQyAtpr/haaJeM=
-golang.org/x/mod v0.35.0/go.mod h1:+GwiRhIInF8wPm+4AoT6L0FA1QWAad3OMdTRx4tFYlU=
+golang.org/x/image v0.41.0 h1:8wS72eGJMJaBxK6okTzd4WaXumUlTVlb753MlsSvTCo=
+golang.org/x/image v0.41.0/go.mod h1:uIc348UZMSvS5Z65CVZ7iDPaNobNFEPeJ4kbqTOszmA=
 golang.org/x/mod v0.37.0 h1:vF1DjpVEshcIqoEaauuHebaLk1O1forxjxBaVn884JQ=
 golang.org/x/mod v0.37.0/go.mod h1:m8S8VeM9r4dzDwjrKO0a1sZP3YjeMamRRlD+fmR2Q/0=
-golang.org/x/net v0.55.0 h1:bcvxaJn3e1U6InsFWt1JUq1aSjnRxLzT2rtD2KfkDF8=
-golang.org/x/net v0.55.0/go.mod h1:L5U2KuzuOe1lY7Z+aWVIKK6qEeJXnXV9yzGA+WCHJww=
 golang.org/x/net v0.56.0 h1:Rw8j/hFzGvJUZwNBXnAtf5sVDVt+65SK2C7IxCxZt5o=
 golang.org/x/net v0.56.0/go.mod h1:D3Ku6r+V6JROoZK144D2XfMHFcMq/0zSfLelVTCFKec=
-golang.org/x/sync v0.20.0 h1:e0PTpb7pjO8GAtTs2dQ6jYa5BWYlMuX047Dco/pItO4=
-golang.org/x/sync v0.20.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
 golang.org/x/sync v0.21.0 h1:HLII4xRRTtCRkxYp4HNFF0Js/Og6q2i++KXbg0gHCwM=
 golang.org/x/sync v0.21.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
 golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
@@ -448,22 +450,14 @@ golang.org/x/sys v0.0.0-20220811171246-fbc7d0a398ab/go.mod h1:oPkhp1MJrh7nUepCBc
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.8.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.11.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.45.0 h1:dO4czNzziLiiXplLQgBCEpCvXQ3dnkn0SdaZSYdQ+FY=
-golang.org/x/sys v0.45.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
 golang.org/x/sys v0.46.0 h1:noSf2Fq6F8DBgS+LysIkx7rIExoNHJsxOAtPp4rthXw=
 golang.org/x/sys v0.46.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
-golang.org/x/term v0.43.0 h1:S4RLU2sB31O/NCl+zFN9Aru9A/Cq2aqKpTZJ6B+DwT4=
-golang.org/x/term v0.43.0/go.mod h1:lrhlHNdQJHO+1qVYiHfFKVuVioJIheAc3fBSMFYEIsk=
 golang.org/x/term v0.44.0 h1:0rLvDRCtNj0gZkyIXhCyOb2OAzEhLVqc4B+hrsBhrmc=
 golang.org/x/term v0.44.0/go.mod h1:7ze4MdzUzLXpSAoFP1H0bOI9aXDqveSvatT5vKcFh2Y=
-golang.org/x/text v0.37.0 h1:Cqjiwd9eSg8e0QAkyCaQTNHFIIzWtidPahFWR83rTrc=
-golang.org/x/text v0.37.0/go.mod h1:a5sjxXGs9hsn/AJVwuElvCAo9v8QYLzvavO5z2PiM38=
 golang.org/x/text v0.39.0 h1:UbZz4pLOvn600D6Oh6GGEI6VAmndrEBLv8/6BEXzyus=
 golang.org/x/text v0.39.0/go.mod h1:3UwRclnC2g0TU9x8PZiyfOajCd1zaUNHF9cvqcQZ+ZM=
 golang.org/x/time v0.12.0 h1:ScB/8o8olJvc+CQPWrK3fPZNfh7qgwCrY0zJmoEQLSE=
 golang.org/x/time v0.12.0/go.mod h1:CDIdPxbZBQxdj6cxyCIdrNogrJKMJ7pr37NYpMcMDSg=
-golang.org/x/tools v0.44.0 h1:UP4ajHPIcuMjT1GqzDWRlalUEoY+uzoZKnhOjbIPD2c=
-golang.org/x/tools v0.44.0/go.mod h1:KA0AfVErSdxRZIsOVipbv3rQhVXTnlU6UhKxHd1seDI=
 golang.org/x/tools v0.47.0 h1:7Kn5x/d1svx/PzryTsqeoZN4TZwqeH5pGWjefhLi/1Q=
 golang.org/x/tools v0.47.0/go.mod h1:dFHnyTvFWY212G+h7ZY4Vsp/K3U4/7W9TyVaAul8uCA=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=

backend/internal/config/config.go

@@ -73,6 +73,7 @@ type Config struct {
 	Ops                     OpsConfig                     `mapstructure:"ops"`
 	JWT                     JWTConfig                     `mapstructure:"jwt"`
 	Totp                    TotpConfig                    `mapstructure:"totp"`
+	WebAuthn                WebAuthnConfig                `mapstructure:"webauthn"`
 	LinuxDo                 LinuxDoConnectConfig          `mapstructure:"linuxdo_connect"`
 	WeChat                  WeChatConnectConfig           `mapstructure:"wechat_connect"`
 	OIDC                    OIDCConnectConfig             `mapstructure:"oidc_connect"`
@@ -686,6 +687,16 @@ type CORSConfig struct {
 	AllowCredentials bool     `mapstructure:"allow_credentials"`
 }
 
+// WebAuthnConfig configures this deployment as a WebAuthn relying party.
+// RPID and RPOrigins are security boundaries and must never be inferred from
+// untrusted request Host or Origin headers.
+type WebAuthnConfig struct {
+	Enabled       bool     `mapstructure:"enabled"`
+	RPDisplayName string   `mapstructure:"rp_display_name"`
+	RPID          string   `mapstructure:"rp_id"`
+	RPOrigins     []string `mapstructure:"rp_origins"`
+}
+
 const MaxForwardedClientIPHeaders = 16
 
 type ForwardedClientIPSettings struct {
@@ -913,6 +924,8 @@ type GatewayConfig struct {
 	OpenAICompactModel string `mapstructure:"openai_compact_model"`
 	// OpenAIWS: OpenAI Responses WebSocket 配置（默认开启，可按需回滚到 HTTP）
 	OpenAIWS GatewayOpenAIWSConfig `mapstructure:"openai_ws"`
+	// Live: ChatGPT Frameless Live 会话配置。
+	Live GatewayLiveConfig `mapstructure:"live"`
 	// OpenAIScheduler: OpenAI 高级调度器粘性逃逸配置
 	OpenAIScheduler GatewayOpenAISchedulerConfig `mapstructure:"openai_scheduler"`
 	// OpenAIHTTP2: OpenAI HTTP 上游协议策略（默认启用 HTTP/2，可按代理能力回退 HTTP/1.1）
@@ -999,6 +1012,11 @@ type GatewayConfig struct {
 	UserMessageQueue UserMessageQueueConfig `mapstructure:"user_message_queue"`
 }
 
+type GatewayLiveConfig struct {
+	// MaxSessionDurationSeconds 是 Live 会话的硬上限。
+	MaxSessionDurationSeconds int `mapstructure:"max_session_duration_seconds"`
+}
+
 // GatewayOpenAIHTTP2Config OpenAI HTTP 上游协议配置。
 // 默认启用 HTTP/2；在部分代理不兼容时按策略回退 HTTP/1.1。
 type GatewayOpenAIHTTP2Config struct {
@@ -1017,6 +1035,8 @@ type GatewayOpenAIHTTP2Config struct {
 // GatewayOpenAIProxyStreamCircuitConfig controls the bounded, in-process
 // proxy-ID circuit used for incomplete OpenAI Responses SSE streams.
 type GatewayOpenAIProxyStreamCircuitConfig struct {
+	// Disabled: 完全关闭代理断流熔断（默认开启）。
+	Disabled bool `mapstructure:"disabled"`
 	// FailureThreshold: 统计窗口内多少次断流后隔离代理。
 	FailureThreshold int `mapstructure:"failure_threshold"`
 	// WindowSeconds: 断流统计窗口（秒）。
@@ -1633,20 +1653,7 @@ func LoadForBootstrap() (*Config, error) {
 func load(allowMissingJWTSecret bool) (*Config, error) {
 	viper.SetConfigName("config")
 	viper.SetConfigType("yaml")
-
-	// Add config paths in priority order
-	// 1. DATA_DIR environment variable (highest priority)
-	if dataDir := os.Getenv("DATA_DIR"); dataDir != "" {
-		viper.AddConfigPath(dataDir)
-	}
-	// 2. Docker data directory
-	viper.AddConfigPath("/app/data")
-	// 3. Current directory
-	viper.AddConfigPath(".")
-	// 4. Config subdirectory
-	viper.AddConfigPath("./config")
-	// 5. System config directory
-	viper.AddConfigPath("/etc/sub2api")
+	configureConfigSource(viper.SetConfigFile, viper.AddConfigPath)
 
 	// 环境变量支持
 	viper.AutomaticEnv()
@@ -1821,6 +1828,22 @@ func load(allowMissingJWTSecret bool) (*Config, error) {
 	return &cfg, nil
 }
 
+func configureConfigSource(setConfigFile, addConfigPath func(string)) {
+	if configFile := strings.TrimSpace(os.Getenv("CONFIG_FILE")); configFile != "" {
+		setConfigFile(configFile)
+		return
+	}
+
+	// Add config paths in priority order.
+	if dataDir := strings.TrimSpace(os.Getenv("DATA_DIR")); dataDir != "" {
+		addConfigPath(dataDir)
+	}
+	addConfigPath("/app/data")
+	addConfigPath(".")
+	addConfigPath("./config")
+	addConfigPath("/etc/sub2api")
+}
+
 func setDefaults() {
 	viper.SetDefault("run_mode", RunModeStandard)
 
@@ -1865,12 +1888,21 @@ func setDefaults() {
 	viper.SetDefault("cors.allowed_origins", []string{})
 	viper.SetDefault("cors.allow_credentials", true)
 
+	// WebAuthn / Passkeys are opt-in because every deployment must explicitly
+	// declare its relying-party domain and trusted browser origins.
+	viper.SetDefault("webauthn.enabled", false)
+	viper.SetDefault("webauthn.rp_display_name", "Sub2API")
+	viper.SetDefault("webauthn.rp_id", "")
+	viper.SetDefault("webauthn.rp_origins", []string{})
+
 	// Security
 	viper.SetDefault("security.url_allowlist.enabled", false)
 	viper.SetDefault("security.url_allowlist.upstream_hosts", []string{
 		"api.openai.com",
 		"api.anthropic.com",
 		"api.kimi.com",
+		"api.moonshot.ai",
+		"api.moonshot.cn",
 		"open.bigmodel.cn",
 		"api.minimaxi.com",
 		"generativelanguage.googleapis.com",
@@ -2188,6 +2220,7 @@ func setDefaults() {
 	viper.SetDefault("gateway.codex_image_generation_bridge_enabled", false)
 	viper.SetDefault("gateway.openai_passthrough_allow_timeout_headers", false)
 	viper.SetDefault("gateway.openai_compact_model", "gpt-5.4")
+	viper.SetDefault("gateway.live.max_session_duration_seconds", 3600)
 	// OpenAI Responses WebSocket（默认开启；可通过 force_http 紧急回滚）
 	viper.SetDefault("gateway.openai_ws.enabled", true)
 	viper.SetDefault("gateway.openai_ws.mode_router_v2_enabled", false)
@@ -2251,6 +2284,7 @@ func setDefaults() {
 	viper.SetDefault("gateway.openai_http2.fallback_error_threshold", 2)
 	viper.SetDefault("gateway.openai_http2.fallback_window_seconds", 60)
 	viper.SetDefault("gateway.openai_http2.fallback_ttl_seconds", 600)
+	viper.SetDefault("gateway.openai_proxy_stream_circuit.disabled", false)
 	viper.SetDefault("gateway.openai_proxy_stream_circuit.failure_threshold", 2)
 	viper.SetDefault("gateway.openai_proxy_stream_circuit.window_seconds", 60)
 	viper.SetDefault("gateway.openai_proxy_stream_circuit.ttl_seconds", 600)
@@ -2580,6 +2614,43 @@ func (c *Config) Validate() error {
 		}
 		warnIfInsecureURL("server.frontend_url", c.Server.FrontendURL)
 	}
+	if c.WebAuthn.Enabled {
+		c.WebAuthn.RPDisplayName = strings.TrimSpace(c.WebAuthn.RPDisplayName)
+		c.WebAuthn.RPID = strings.ToLower(strings.TrimSpace(c.WebAuthn.RPID))
+		c.WebAuthn.RPOrigins = normalizeStringSlice(c.WebAuthn.RPOrigins)
+		if c.WebAuthn.RPDisplayName == "" {
+			return fmt.Errorf("webauthn.rp_display_name is required when passkeys are enabled")
+		}
+		if c.WebAuthn.RPID == "" {
+			return fmt.Errorf("webauthn.rp_id is required when passkeys are enabled")
+		}
+		if strings.Contains(c.WebAuthn.RPID, "://") || strings.ContainsAny(c.WebAuthn.RPID, "/:") {
+			return fmt.Errorf("webauthn.rp_id must be a domain without scheme, port, or path")
+		}
+		if len(c.WebAuthn.RPOrigins) == 0 {
+			return fmt.Errorf("webauthn.rp_origins must contain at least one origin when passkeys are enabled")
+		}
+		for i, origin := range c.WebAuthn.RPOrigins {
+			u, err := url.Parse(origin)
+			if err != nil || u.Scheme == "" || u.Host == "" {
+				return fmt.Errorf("webauthn.rp_origins contains invalid origin %q", origin)
+			}
+			if u.User != nil || u.RawQuery != "" || u.Fragment != "" || u.Path != "" {
+				return fmt.Errorf("webauthn.rp_origins entry %q must not include userinfo, path, query, or fragment", origin)
+			}
+			u.Scheme = strings.ToLower(u.Scheme)
+			u.Host = strings.ToLower(u.Host)
+			host := strings.ToLower(u.Hostname())
+			localDevelopment := host == "localhost" || host == "127.0.0.1" || host == "::1"
+			if u.Scheme != "https" && (u.Scheme != "http" || !localDevelopment) {
+				return fmt.Errorf("webauthn.rp_origins entry %q must use HTTPS (HTTP is allowed only for localhost)", origin)
+			}
+			if host != c.WebAuthn.RPID && !strings.HasSuffix(host, "."+c.WebAuthn.RPID) {
+				return fmt.Errorf("webauthn.rp_origins entry %q is not within relying party ID %q", origin, c.WebAuthn.RPID)
+			}
+			c.WebAuthn.RPOrigins[i] = u.Scheme + "://" + u.Host
+		}
+	}
 	if c.JWT.ExpireHour <= 0 {
 		return fmt.Errorf("jwt.expire_hour must be positive")
 	}
@@ -3039,6 +3110,9 @@ func (c *Config) Validate() error {
 		(c.Gateway.OpenAIHighEffortFirstOutputTimeoutSeconds > 0 && c.Gateway.OpenAIHighEffortFirstOutputTimeoutSeconds < 30) {
 		return fmt.Errorf("gateway.openai_high_effort_first_output_timeout_seconds must be 0 or between 30-1800 seconds")
 	}
+	if c.Gateway.Live.MaxSessionDurationSeconds <= 0 {
+		c.Gateway.Live.MaxSessionDurationSeconds = 3600
+	}
 	if strings.TrimSpace(c.Gateway.ConnectionPoolIsolation) != "" {
 		switch c.Gateway.ConnectionPoolIsolation {
 		case ConnectionPoolIsolationProxy, ConnectionPoolIsolationAccount, ConnectionPoolIsolationAccountProxy:
@@ -3477,14 +3551,12 @@ func generateJWTSecret(byteLength int) (string, error) {
 // GetServerAddress returns the server address (host:port) from config file or environment variable.
 // This is a lightweight function that can be used before full config validation,
 // such as during setup wizard startup.
-// Priority: config.yaml > environment variables > defaults
+// Priority: environment variables > config file > defaults.
 func GetServerAddress() string {
 	v := viper.New()
 	v.SetConfigName("config")
 	v.SetConfigType("yaml")
-	v.AddConfigPath(".")
-	v.AddConfigPath("./config")
-	v.AddConfigPath("/etc/sub2api")
+	configureConfigSource(v.SetConfigFile, v.AddConfigPath)
 
 	// Support SERVER_HOST and SERVER_PORT environment variables
 	v.AutomaticEnv()

backend/internal/config/config_test.go

@@ -17,6 +17,9 @@ import (
 func resetViperWithJWTSecret(t *testing.T) {
 	t.Helper()
 	viper.Reset()
+	t.Cleanup(viper.Reset)
+	t.Setenv("CONFIG_FILE", "")
+	t.Setenv("DATA_DIR", "")
 	t.Setenv("JWT_SECRET", strings.Repeat("x", 32))
 }
 
@@ -253,8 +256,53 @@ func TestLoadTrustedProxiesPresenceFromYAML(t *testing.T) {
 	}
 }
 
+func TestLoadTrustedProxiesFromConfigFile(t *testing.T) {
+	resetViperWithJWTSecret(t)
+	configFile := filepath.Join(t.TempDir(), "reporter-config.yaml")
+	require.NoError(t, os.WriteFile(configFile, []byte(`server:
+  host: 192.0.2.10
+  trusted_proxies:
+    - 127.0.0.1/32
+    - 10.0.0.0/8
+    - 172.16.0.0/12
+    - 192.168.0.0/16
+`), 0o600))
+	t.Setenv("CONFIG_FILE", configFile)
+
+	cfg, err := Load()
+	require.NoError(t, err)
+	require.Equal(t, "192.0.2.10", cfg.Server.Host)
+	require.Equal(t, []string{"127.0.0.1/32", "10.0.0.0/8", "172.16.0.0/12", "192.168.0.0/16"}, cfg.Server.TrustedProxies)
+	require.True(t, cfg.Server.TrustedProxiesConfigured)
+}
+
+func TestConfigFileTakesPrecedenceOverDataDir(t *testing.T) {
+	resetViperWithJWTSecret(t)
+	dataDir := t.TempDir()
+	require.NoError(t, os.WriteFile(filepath.Join(dataDir, "config.yaml"), []byte("server:\n  host: 192.0.2.20\n"), 0o600))
+	configFile := filepath.Join(t.TempDir(), "explicit.yaml")
+	require.NoError(t, os.WriteFile(configFile, []byte("server:\n  host: 192.0.2.30\n"), 0o600))
+	t.Setenv("DATA_DIR", dataDir)
+	t.Setenv("CONFIG_FILE", configFile)
+
+	cfg, err := Load()
+	require.NoError(t, err)
+	require.Equal(t, "192.0.2.30", cfg.Server.Host)
+}
+
+func TestLoadReturnsErrorForMissingConfigFile(t *testing.T) {
+	resetViperWithJWTSecret(t)
+	t.Setenv("CONFIG_FILE", filepath.Join(t.TempDir(), "missing.yaml"))
+
+	_, err := Load()
+	require.ErrorContains(t, err, "read config error")
+}
+
 func TestLoadForBootstrapAllowsMissingJWTSecret(t *testing.T) {
 	viper.Reset()
+	t.Cleanup(viper.Reset)
+	t.Setenv("CONFIG_FILE", "")
+	t.Setenv("DATA_DIR", "")
 	t.Setenv("JWT_SECRET", "")
 
 	cfg, err := LoadForBootstrap()
@@ -497,6 +545,7 @@ func TestLoadDefaultOpenAIHTTP2Enabled(t *testing.T) {
 	require.NoError(t, err)
 	require.True(t, cfg.Gateway.OpenAIHTTP2.Enabled)
 	require.True(t, cfg.Gateway.OpenAIHTTP2.AllowProxyFallbackToHTTP1)
+	require.False(t, cfg.Gateway.OpenAIProxyStreamCircuit.Disabled)
 	require.Equal(t, 2, cfg.Gateway.OpenAIProxyStreamCircuit.FailureThreshold)
 	require.Equal(t, 60, cfg.Gateway.OpenAIProxyStreamCircuit.WindowSeconds)
 	require.Equal(t, 600, cfg.Gateway.OpenAIProxyStreamCircuit.TTLSeconds)
@@ -504,12 +553,14 @@ func TestLoadDefaultOpenAIHTTP2Enabled(t *testing.T) {
 
 func TestLoadOpenAIProxyStreamCircuitFromEnv(t *testing.T) {
 	resetViperWithJWTSecret(t)
+	t.Setenv("GATEWAY_OPENAI_PROXY_STREAM_CIRCUIT_DISABLED", "true")
 	t.Setenv("GATEWAY_OPENAI_PROXY_STREAM_CIRCUIT_FAILURE_THRESHOLD", "3")
 	t.Setenv("GATEWAY_OPENAI_PROXY_STREAM_CIRCUIT_WINDOW_SECONDS", "90")
 	t.Setenv("GATEWAY_OPENAI_PROXY_STREAM_CIRCUIT_TTL_SECONDS", "420")
 
 	cfg, err := Load()
 	require.NoError(t, err)
+	require.True(t, cfg.Gateway.OpenAIProxyStreamCircuit.Disabled)
 	require.Equal(t, 3, cfg.Gateway.OpenAIProxyStreamCircuit.FailureThreshold)
 	require.Equal(t, 90, cfg.Gateway.OpenAIProxyStreamCircuit.WindowSeconds)
 	require.Equal(t, 420, cfg.Gateway.OpenAIProxyStreamCircuit.TTLSeconds)
@@ -707,6 +758,21 @@ func TestLoadDefaultSecurityToggles(t *testing.T) {
 	if !cfg.Security.ResponseHeaders.Enabled {
 		t.Fatalf("ResponseHeaders.Enabled = false, want true")
 	}
+
+	wantHosts := []string{
+		"api.kimi.com",
+		"api.moonshot.ai",
+		"api.moonshot.cn",
+	}
+	hostSet := make(map[string]struct{}, len(cfg.Security.URLAllowlist.UpstreamHosts))
+	for _, h := range cfg.Security.URLAllowlist.UpstreamHosts {
+		hostSet[h] = struct{}{}
+	}
+	for _, want := range wantHosts {
+		if _, ok := hostSet[want]; !ok {
+			t.Fatalf("URLAllowlist.UpstreamHosts missing %q; got %v", want, cfg.Security.URLAllowlist.UpstreamHosts)
+		}
+	}
 }
 
 func TestLoadDefaultServerMode(t *testing.T) {
@@ -1155,6 +1221,8 @@ func TestNormalizeStringSlice(t *testing.T) {
 }
 
 func TestGetServerAddressFromEnv(t *testing.T) {
+	t.Setenv("CONFIG_FILE", "")
+	t.Setenv("DATA_DIR", "")
 	t.Setenv("SERVER_HOST", "127.0.0.1")
 	t.Setenv("SERVER_PORT", "9090")
 
@@ -1164,6 +1232,17 @@ func TestGetServerAddressFromEnv(t *testing.T) {
 	}
 }
 
+func TestGetServerAddressFromConfigFile(t *testing.T) {
+	configFile := filepath.Join(t.TempDir(), "setup.yaml")
+	require.NoError(t, os.WriteFile(configFile, []byte("server:\n  host: 192.0.2.40\n  port: 9191\n"), 0o600))
+	t.Setenv("CONFIG_FILE", configFile)
+	t.Setenv("DATA_DIR", "")
+	t.Setenv("SERVER_HOST", "")
+	t.Setenv("SERVER_PORT", "")
+
+	require.Equal(t, "192.0.2.40:9191", GetServerAddress())
+}
+
 func TestValidateAbsoluteHTTPURL(t *testing.T) {
 	if err := ValidateAbsoluteHTTPURL("https://example.com/path"); err != nil {
 		t.Fatalf("ValidateAbsoluteHTTPURL valid url error: %v", err)

backend/internal/config/webauthn_test.go

@@ -0,0 +1,104 @@
+package config
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/spf13/viper"
+	"github.com/stretchr/testify/require"
+)
+
+func TestValidateWebAuthnConfig(t *testing.T) {
+	tests := []struct {
+		name      string
+		configure func(*Config)
+		wantError string
+	}{
+		{
+			name: "valid production origin",
+			configure: func(cfg *Config) {
+				cfg.WebAuthn = WebAuthnConfig{
+					Enabled:       true,
+					RPDisplayName: "Sub2API",
+					RPID:          "sub2api.example.com",
+					RPOrigins:     []string{"https://sub2api.example.com"},
+				}
+			},
+		},
+		{
+			name: "valid localhost development origin",
+			configure: func(cfg *Config) {
+				cfg.WebAuthn = WebAuthnConfig{
+					Enabled:       true,
+					RPDisplayName: "Sub2API Dev",
+					RPID:          "localhost",
+					RPOrigins:     []string{"http://localhost:5173"},
+				}
+			},
+		},
+		{
+			name: "missing relying party id",
+			configure: func(cfg *Config) {
+				cfg.WebAuthn = WebAuthnConfig{
+					Enabled:       true,
+					RPDisplayName: "Sub2API",
+					RPOrigins:     []string{"https://sub2api.example.com"},
+				}
+			},
+			wantError: "webauthn.rp_id",
+		},
+		{
+			name: "relying party id contains scheme",
+			configure: func(cfg *Config) {
+				cfg.WebAuthn = WebAuthnConfig{
+					Enabled:       true,
+					RPDisplayName: "Sub2API",
+					RPID:          "https://sub2api.example.com",
+					RPOrigins:     []string{"https://sub2api.example.com"},
+				}
+			},
+			wantError: "domain without scheme",
+		},
+		{
+			name: "non-local insecure origin",
+			configure: func(cfg *Config) {
+				cfg.WebAuthn = WebAuthnConfig{
+					Enabled:       true,
+					RPDisplayName: "Sub2API",
+					RPID:          "sub2api.example.com",
+					RPOrigins:     []string{"http://sub2api.example.com"},
+				}
+			},
+			wantError: "must use HTTPS",
+		},
+		{
+			name: "origin outside relying party id",
+			configure: func(cfg *Config) {
+				cfg.WebAuthn = WebAuthnConfig{
+					Enabled:       true,
+					RPDisplayName: "Sub2API",
+					RPID:          "example.com",
+					RPOrigins:     []string{"https://example.net"},
+				}
+			},
+			wantError: "not within relying party ID",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			viper.Reset()
+			t.Setenv("JWT_SECRET", strings.Repeat("x", 32))
+			cfg, err := Load()
+			require.NoError(t, err)
+			tt.configure(cfg)
+
+			err = cfg.Validate()
+			if tt.wantError == "" {
+				require.NoError(t, err)
+			} else {
+				require.ErrorContains(t, err, tt.wantError)
+			}
+		})
+	}
+}

backend/internal/domain/constants.go

@@ -133,6 +133,7 @@ var DefaultBedrockModelMapping = map[string]string{
 	// Claude Fable
 	"claude-fable-5": "anthropic.claude-fable-5",
 	// Claude Opus
+	"claude-opus-5":            "us.anthropic.claude-opus-5-v1",
 	"claude-opus-4-8":          "us.anthropic.claude-opus-4-8-v1",
 	"claude-opus-4-7":          "us.anthropic.claude-opus-4-7-v1",
 	"claude-opus-4-6-thinking": "us.anthropic.claude-opus-4-6-v1",

backend/internal/handler/admin/account_handler.go

@@ -147,6 +147,8 @@ type UpdateAccountRequest struct {
 	GroupIDs                *[]int64       `json:"group_ids"`
 	ExpiresAt               *int64         `json:"expires_at"`
 	AutoPauseOnExpired      *bool          `json:"auto_pause_on_expired"`
+	ProbeEnabled            *bool          `json:"upstream_billing_probe_enabled"`
+	RateSyncEnabled         *bool          `json:"upstream_billing_rate_sync_enabled"`
 	ConfirmMixedChannelRisk *bool          `json:"confirm_mixed_channel_risk"` // 用户确认混合渠道风险
 }
 
@@ -987,6 +989,8 @@ func (h *AccountHandler) Update(c *gin.Context) {
 		GroupIDs:              req.GroupIDs,
 		ExpiresAt:             req.ExpiresAt,
 		AutoPauseOnExpired:    req.AutoPauseOnExpired,
+		ProbeEnabled:          req.ProbeEnabled,
+		RateSyncEnabled:       req.RateSyncEnabled,
 		SkipMixedChannelCheck: skipCheck,
 	})
 	if err != nil {
@@ -1531,6 +1535,141 @@ func (h *AccountHandler) RevertProxyFallback(c *gin.Context) {
 	response.Success(c, gin.H{"message": "reverted"})
 }
 
+// BatchDelete handles deleting multiple accounts with bounded concurrency.
+// POST /api/v1/admin/accounts/batch-delete
+func (h *AccountHandler) BatchDelete(c *gin.Context) {
+	var req struct {
+		AccountIDs []int64 `json:"account_ids"`
+	}
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.BadRequest(c, "Invalid request: "+err.Error())
+		return
+	}
+
+	accountIDs := normalizeInt64IDList(req.AccountIDs)
+	if len(accountIDs) == 0 {
+		response.BadRequest(c, "account_ids is required")
+		return
+	}
+
+	accounts, err := h.adminService.GetAccountsByIDs(c.Request.Context(), accountIDs)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	type deleteError struct {
+		AccountID int64  `json:"account_id"`
+		Error     string `json:"error"`
+	}
+
+	requestedIDs := make(map[int64]struct{}, len(accountIDs))
+	for _, accountID := range accountIDs {
+		requestedIDs[accountID] = struct{}{}
+	}
+	accountsByID := make(map[int64]*service.Account, len(accounts))
+	for _, account := range accounts {
+		if account != nil {
+			accountsByID[account.ID] = account
+		}
+	}
+
+	rootIDs := make([]int64, 0, len(accountIDs))
+	dependentIDs := make(map[int64][]int64)
+	failedIDs := make([]int64, 0)
+	errorsByAccount := make([]deleteError, 0)
+	for _, accountID := range accountIDs {
+		account := accountsByID[accountID]
+		if account == nil {
+			failedIDs = append(failedIDs, accountID)
+			errorsByAccount = append(errorsByAccount, deleteError{
+				AccountID: accountID,
+				Error:     "account not found",
+			})
+			continue
+		}
+
+		rootID := accountID
+		visited := map[int64]struct{}{accountID: {}}
+		for {
+			current := accountsByID[rootID]
+			if current == nil || current.ParentAccountID == nil {
+				break
+			}
+			parentID := *current.ParentAccountID
+			if _, selected := requestedIDs[parentID]; !selected {
+				break
+			}
+			if _, exists := accountsByID[parentID]; !exists {
+				break
+			}
+			if _, cyclic := visited[parentID]; cyclic {
+				rootID = accountID
+				break
+			}
+			visited[parentID] = struct{}{}
+			rootID = parentID
+		}
+
+		if rootID != accountID {
+			dependentIDs[rootID] = append(dependentIDs[rootID], accountID)
+			continue
+		}
+		rootIDs = append(rootIDs, accountID)
+	}
+
+	const maxConcurrency = 5
+	g, gctx := errgroup.WithContext(c.Request.Context())
+	g.SetLimit(maxConcurrency)
+
+	var mu sync.Mutex
+	successIDs := make([]int64, 0, len(accountIDs))
+
+	// Every worker returns nil so one account failure does not cancel the remaining deletions.
+	for _, id := range rootIDs {
+		accountID := id
+		g.Go(func() error {
+			err := h.adminService.DeleteAccount(gctx, accountID)
+
+			mu.Lock()
+			defer mu.Unlock()
+			affectedIDs := append([]int64{accountID}, dependentIDs[accountID]...)
+			if err != nil {
+				for _, affectedID := range affectedIDs {
+					failedIDs = append(failedIDs, affectedID)
+					errorsByAccount = append(errorsByAccount, deleteError{
+						AccountID: affectedID,
+						Error:     err.Error(),
+					})
+				}
+				return nil
+			}
+			successIDs = append(successIDs, affectedIDs...)
+			return nil
+		})
+	}
+
+	if err := g.Wait(); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	sort.Slice(successIDs, func(i, j int) bool { return successIDs[i] < successIDs[j] })
+	sort.Slice(failedIDs, func(i, j int) bool { return failedIDs[i] < failedIDs[j] })
+	sort.Slice(errorsByAccount, func(i, j int) bool {
+		return errorsByAccount[i].AccountID < errorsByAccount[j].AccountID
+	})
+
+	response.Success(c, gin.H{
+		"total":       len(accountIDs),
+		"success":     len(successIDs),
+		"failed":      len(failedIDs),
+		"success_ids": successIDs,
+		"failed_ids":  failedIDs,
+		"errors":      errorsByAccount,
+	})
+}
+
 // BatchClearError handles batch clearing account errors
 // POST /api/v1/admin/accounts/batch-clear-error
 func (h *AccountHandler) BatchClearError(c *gin.Context) {

backend/internal/handler/admin/account_handler_batch_delete_test.go

@@ -0,0 +1,173 @@
+package admin
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"errors"
+	"net/http"
+	"net/http/httptest"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/require"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+)
+
+type batchDeleteAdminService struct {
+	*stubAdminService
+
+	mu               sync.Mutex
+	active           int
+	maxActive        int
+	deletedIDs       []int64
+	deleteErrorsByID map[int64]error
+	accountsByID     map[int64]*service.Account
+}
+
+func (s *batchDeleteAdminService) GetAccountsByIDs(_ context.Context, ids []int64) ([]*service.Account, error) {
+	accounts := make([]*service.Account, 0, len(ids))
+	for _, id := range ids {
+		if s.accountsByID != nil {
+			if account, ok := s.accountsByID[id]; ok {
+				accounts = append(accounts, account)
+			}
+			continue
+		}
+		accounts = append(accounts, &service.Account{ID: id})
+	}
+	return accounts, nil
+}
+
+func (s *batchDeleteAdminService) DeleteAccount(ctx context.Context, id int64) error {
+	s.mu.Lock()
+	s.active++
+	if s.active > s.maxActive {
+		s.maxActive = s.active
+	}
+	s.mu.Unlock()
+
+	select {
+	case <-ctx.Done():
+		s.mu.Lock()
+		s.active--
+		s.mu.Unlock()
+		return ctx.Err()
+	case <-time.After(10 * time.Millisecond):
+	}
+
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	s.active--
+	s.deletedIDs = append(s.deletedIDs, id)
+	return s.deleteErrorsByID[id]
+}
+
+func setupAccountBatchDeleteRouter(adminSvc *batchDeleteAdminService) *gin.Engine {
+	gin.SetMode(gin.TestMode)
+	router := gin.New()
+	handler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	router.POST("/api/v1/admin/accounts/batch-delete", handler.BatchDelete)
+	return router
+}
+
+func TestAccountHandlerBatchDeleteReturnsStablePerAccountResults(t *testing.T) {
+	adminSvc := &batchDeleteAdminService{
+		stubAdminService: newStubAdminService(),
+		deleteErrorsByID: map[int64]error{
+			3: errors.New("delete failed"),
+		},
+	}
+	router := setupAccountBatchDeleteRouter(adminSvc)
+
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequest(
+		http.MethodPost,
+		"/api/v1/admin/accounts/batch-delete",
+		bytes.NewBufferString(`{"account_ids":[5,4,3,2,1,2,0,-1]}`),
+	)
+	req.Header.Set("Content-Type", "application/json")
+	router.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusOK, rec.Code)
+
+	var payload struct {
+		Data struct {
+			Total      int     `json:"total"`
+			Success    int     `json:"success"`
+			Failed     int     `json:"failed"`
+			SuccessIDs []int64 `json:"success_ids"`
+			FailedIDs  []int64 `json:"failed_ids"`
+			Errors     []struct {
+				AccountID int64  `json:"account_id"`
+				Error     string `json:"error"`
+			} `json:"errors"`
+		} `json:"data"`
+	}
+	require.NoError(t, json.Unmarshal(rec.Body.Bytes(), &payload))
+	require.Equal(t, 5, payload.Data.Total)
+	require.Equal(t, 4, payload.Data.Success)
+	require.Equal(t, 1, payload.Data.Failed)
+	require.Equal(t, []int64{1, 2, 4, 5}, payload.Data.SuccessIDs)
+	require.Equal(t, []int64{3}, payload.Data.FailedIDs)
+	require.Equal(t, int64(3), payload.Data.Errors[0].AccountID)
+	require.Equal(t, "delete failed", payload.Data.Errors[0].Error)
+	require.LessOrEqual(t, adminSvc.maxActive, 5)
+	require.Greater(t, adminSvc.maxActive, 1)
+}
+
+func TestAccountHandlerBatchDeleteDoesNotRaceSelectedShadowWithParent(t *testing.T) {
+	parentID := int64(1)
+	adminSvc := &batchDeleteAdminService{
+		stubAdminService: newStubAdminService(),
+		accountsByID: map[int64]*service.Account{
+			1: {ID: 1},
+			2: {ID: 2, ParentAccountID: &parentID},
+			3: {ID: 3},
+		},
+	}
+	router := setupAccountBatchDeleteRouter(adminSvc)
+
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequest(
+		http.MethodPost,
+		"/api/v1/admin/accounts/batch-delete",
+		bytes.NewBufferString(`{"account_ids":[1,2,3]}`),
+	)
+	req.Header.Set("Content-Type", "application/json")
+	router.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusOK, rec.Code)
+
+	var payload struct {
+		Data struct {
+			SuccessIDs []int64 `json:"success_ids"`
+			FailedIDs  []int64 `json:"failed_ids"`
+		} `json:"data"`
+	}
+	require.NoError(t, json.Unmarshal(rec.Body.Bytes(), &payload))
+	require.Equal(t, []int64{1, 2, 3}, payload.Data.SuccessIDs)
+	require.Empty(t, payload.Data.FailedIDs)
+	require.ElementsMatch(t, []int64{1, 3}, adminSvc.deletedIDs)
+}
+
+func TestAccountHandlerBatchDeleteRejectsEmptyNormalizedIDs(t *testing.T) {
+	adminSvc := &batchDeleteAdminService{
+		stubAdminService: newStubAdminService(),
+	}
+	router := setupAccountBatchDeleteRouter(adminSvc)
+
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequest(
+		http.MethodPost,
+		"/api/v1/admin/accounts/batch-delete",
+		bytes.NewBufferString(`{"account_ids":[0,-1]}`),
+	)
+	req.Header.Set("Content-Type", "application/json")
+	router.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusBadRequest, rec.Code)
+}

backend/internal/handler/admin/account_handler_mixed_channel_test.go

@@ -147,6 +147,28 @@ func TestAccountHandlerUpdateMixedChannelConflictSimplifiedResponse(t *testing.T
 	require.False(t, hasRequireConfirmation)
 }
 
+func TestAccountHandlerUpdateMapsUpstreamBillingRateSyncSettings(t *testing.T) {
+	adminSvc := newStubAdminService()
+	router := setupAccountMixedChannelRouter(adminSvc)
+	body, _ := json.Marshal(map[string]any{
+		"name":                               "gemini-key",
+		"upstream_billing_probe_enabled":     true,
+		"upstream_billing_rate_sync_enabled": true,
+	})
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequest(http.MethodPut, "/api/v1/admin/accounts/42", bytes.NewReader(body))
+	req.Header.Set("Content-Type", "application/json")
+
+	router.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.NotNil(t, adminSvc.lastUpdateAccountInput)
+	require.NotNil(t, adminSvc.lastUpdateAccountInput.ProbeEnabled)
+	require.True(t, *adminSvc.lastUpdateAccountInput.ProbeEnabled)
+	require.NotNil(t, adminSvc.lastUpdateAccountInput.RateSyncEnabled)
+	require.True(t, *adminSvc.lastUpdateAccountInput.RateSyncEnabled)
+}
+
 func TestAccountHandlerBulkUpdateMixedChannelConflict(t *testing.T) {
 	adminSvc := newStubAdminService()
 	adminSvc.bulkUpdateAccountErr = &service.MixedChannelError{

backend/internal/handler/admin/account_ollama_cloud_usage_test.go

@@ -61,7 +61,7 @@ func (r *ollamaCloudUsageHandlerTestRepo) UpdateOllamaCloudUsageSnapshot(context
 func (r *ollamaCloudUsageHandlerTestRepo) DisableOllamaCloudUsageAutoRefresh(context.Context, *service.Account) error {
 	return nil
 }
-func (r *ollamaCloudUsageHandlerTestRepo) ListDueOllamaCloudUsageAccounts(context.Context, time.Time, int) ([]service.Account, error) {
+func (r *ollamaCloudUsageHandlerTestRepo) ListDueOllamaCloudUsageAccounts(context.Context, time.Time, time.Duration, time.Duration, int) ([]service.Account, error) {
 	return nil, nil
 }
 
@@ -273,4 +273,5 @@ func TestGetOllamaCloudUsageSettingsHandlerSuccess(t *testing.T) {
 	require.Equal(t, http.StatusOK, recorder.Code)
 	require.Contains(t, recorder.Body.String(), `"enabled":false`)
 	require.Contains(t, recorder.Body.String(), `"interval_minutes":60`)
+	require.Contains(t, recorder.Body.String(), `"debounce_minutes":1`)
 }

backend/internal/handler/admin/admin_service_stub_test.go

@@ -32,6 +32,7 @@ type stubAdminService struct {
 	createAccountErr                    error
 	createSparkShadowErr                error
 	updateAccountErr                    error
+	lastUpdateAccountInput              *service.UpdateAccountInput
 	bulkUpdateAccountErr                error
 	lastBulkUpdateAccountInput          *service.BulkUpdateAccountsInput
 	getAccountResult                    *service.Account
@@ -501,6 +502,7 @@ func (s *stubAdminService) RecoverDuplicateAccount(ctx context.Context, id int64
 
 func (s *stubAdminService) UpdateAccount(ctx context.Context, id int64, input *service.UpdateAccountInput) (*service.Account, error) {
 	s.updateAccountCalls++
+	s.lastUpdateAccountInput = input
 	if s.updateAccountErr != nil {
 		return nil, s.updateAccountErr
 	}

backend/internal/handler/admin/content_moderation_handler.go

@@ -20,10 +20,12 @@ func NewContentModerationHandler(svc *service.ContentModerationService) *Content
 }
 
 type contentModerationConfigRequest struct {
-	Enabled              *bool               `json:"enabled"`
-	Mode                 *string             `json:"mode"`
-	BaseURL              *string             `json:"base_url"`
-	Model                *string             `json:"model"`
+	Enabled *bool   `json:"enabled"`
+	Mode    *string `json:"mode"`
+	BaseURL *string `json:"base_url"`
+	Model   *string `json:"model"`
+	// 审计请求使用的代理服务器：null 不修改；0 清除（直连）；>0 指定代理。
+	ProxyID              *int64              `json:"proxy_id"`
 	APIKey               *string             `json:"api_key"`
 	APIKeys              *[]string           `json:"api_keys"`
 	APIKeysMode          string              `json:"api_keys_mode"`
@@ -60,6 +62,7 @@ type contentModerationAPIKeyTestRequest struct {
 	BaseURL   string   `json:"base_url"`
 	Model     string   `json:"model"`
 	TimeoutMS int      `json:"timeout_ms"`
+	ProxyID   *int64   `json:"proxy_id"`
 	Prompt    string   `json:"prompt"`
 	Images    []string `json:"images"`
 }
@@ -88,6 +91,7 @@ func (h *ContentModerationHandler) UpdateConfig(c *gin.Context) {
 		Mode:                           req.Mode,
 		BaseURL:                        req.BaseURL,
 		Model:                          req.Model,
+		ProxyID:                        req.ProxyID,
 		APIKey:                         req.APIKey,
 		APIKeys:                        req.APIKeys,
 		APIKeysMode:                    req.APIKeysMode,
@@ -134,6 +138,7 @@ func (h *ContentModerationHandler) TestAPIKeys(c *gin.Context) {
 		BaseURL:   req.BaseURL,
 		Model:     req.Model,
 		TimeoutMS: req.TimeoutMS,
+		ProxyID:   req.ProxyID,
 		Prompt:    req.Prompt,
 		Images:    req.Images,
 	})

backend/internal/handler/admin/group_handler.go

@@ -13,6 +13,7 @@ import (
 	infraerrors "github.com/Wei-Shaw/sub2api/internal/pkg/errors"
 	"github.com/Wei-Shaw/sub2api/internal/pkg/response"
 	"github.com/Wei-Shaw/sub2api/internal/pkg/timezone"
+	"github.com/Wei-Shaw/sub2api/internal/platform/liveattestation"
 	"github.com/Wei-Shaw/sub2api/internal/service"
 
 	"github.com/gin-gonic/gin"
@@ -25,6 +26,16 @@ type GroupHandler struct {
 	groupCapacityService *service.GroupCapacityService
 }
 
+// GetLiveCapability 返回当前服务端是否具备生成 Live attestation 的运行环境。
+func (h *GroupHandler) GetLiveCapability(c *gin.Context) {
+	err := liveattestation.NewProvider().Check(c.Request.Context())
+	result := gin.H{"supported": err == nil}
+	if err != nil {
+		result["reason"] = err.Error()
+	}
+	response.Success(c, result)
+}
+
 type optionalLimitField struct {
 	set   bool
 	value *float64
@@ -107,6 +118,9 @@ type CreateGroupRequest struct {
 	PeakStart                       string   `json:"peak_start"`
 	PeakEnd                         string   `json:"peak_end"`
 	PeakRateMultiplier              *float64 `json:"peak_rate_multiplier"`
+	ProfitControlEnabled            bool     `json:"profit_control_enabled"`
+	ProfitMinMargin                 *float64 `json:"profit_min_margin"`
+	ProfitSafetyBuffer              *float64 `json:"profit_safety_buffer"`
 	ImagePrice1K                    *float64 `json:"image_price_1k"`
 	ImagePrice2K                    *float64 `json:"image_price_2k"`
 	ImagePrice4K                    *float64 `json:"image_price_4k"`
@@ -125,6 +139,7 @@ type CreateGroupRequest struct {
 	SupportedModelScopes []string `json:"supported_model_scopes"`
 	// OpenAI Messages 调度配置（仅 openai 平台使用）
 	AllowMessagesDispatch       bool                                      `json:"allow_messages_dispatch"`
+	AllowLive                   bool                                      `json:"allow_live"`
 	RequireOAuthOnly            bool                                      `json:"require_oauth_only"`
 	RequirePrivacySet           bool                                      `json:"require_privacy_set"`
 	DefaultMappedModel          string                                    `json:"default_mapped_model"`
@@ -165,6 +180,9 @@ type UpdateGroupRequest struct {
 	PeakStart                       *string  `json:"peak_start"`
 	PeakEnd                         *string  `json:"peak_end"`
 	PeakRateMultiplier              *float64 `json:"peak_rate_multiplier"`
+	ProfitControlEnabled            *bool    `json:"profit_control_enabled"`
+	ProfitMinMargin                 *float64 `json:"profit_min_margin"`
+	ProfitSafetyBuffer              *float64 `json:"profit_safety_buffer"`
 	ImagePrice1K                    *float64 `json:"image_price_1k"`
 	ImagePrice2K                    *float64 `json:"image_price_2k"`
 	ImagePrice4K                    *float64 `json:"image_price_4k"`
@@ -183,6 +201,7 @@ type UpdateGroupRequest struct {
 	SupportedModelScopes *[]string `json:"supported_model_scopes"`
 	// OpenAI Messages 调度配置（仅 openai 平台使用）
 	AllowMessagesDispatch       *bool                                      `json:"allow_messages_dispatch"`
+	AllowLive                   *bool                                      `json:"allow_live"`
 	RequireOAuthOnly            *bool                                      `json:"require_oauth_only"`
 	RequirePrivacySet           *bool                                      `json:"require_privacy_set"`
 	DefaultMappedModel          *string                                    `json:"default_mapped_model"`
@@ -462,6 +481,13 @@ func (h *GroupHandler) Create(c *gin.Context) {
 		return
 	}
 
+	// platform 是 omitempty：预校验必须用与 CreateGroup 落库一致的归一化平台，
+	// 否则省略 platform 的请求会被误判成「平台不支持利润控制」。
+	if err := service.ValidateProfitControlConfig(service.NormalizeGroupPlatform(req.Platform), req.ProfitControlEnabled, float64ValueOrDefault(req.ProfitMinMargin, 0), float64ValueOrDefault(req.ProfitSafetyBuffer, 0)); err != nil {
+		response.BadRequest(c, err.Error())
+		return
+	}
+
 	group, err := h.adminService.CreateGroup(c.Request.Context(), &service.CreateGroupInput{
 		Name:                            req.Name,
 		Description:                     req.Description,
@@ -484,6 +510,9 @@ func (h *GroupHandler) Create(c *gin.Context) {
 		PeakStart:                       req.PeakStart,
 		PeakEnd:                         req.PeakEnd,
 		PeakRateMultiplier:              req.PeakRateMultiplier,
+		ProfitControlEnabled:            req.ProfitControlEnabled,
+		ProfitMinMargin:                 req.ProfitMinMargin,
+		ProfitSafetyBuffer:              req.ProfitSafetyBuffer,
 		ImagePrice1K:                    req.ImagePrice1K,
 		ImagePrice2K:                    req.ImagePrice2K,
 		ImagePrice4K:                    req.ImagePrice4K,
@@ -499,6 +528,7 @@ func (h *GroupHandler) Create(c *gin.Context) {
 		MCPXMLInject:                    req.MCPXMLInject,
 		SupportedModelScopes:            req.SupportedModelScopes,
 		AllowMessagesDispatch:           req.AllowMessagesDispatch,
+		AllowLive:                       req.AllowLive,
 		RequireOAuthOnly:                req.RequireOAuthOnly,
 		RequirePrivacySet:               req.RequirePrivacySet,
 		DefaultMappedModel:              req.DefaultMappedModel,
@@ -602,6 +632,9 @@ func (h *GroupHandler) Update(c *gin.Context) {
 		PeakStart:                       req.PeakStart,
 		PeakEnd:                         req.PeakEnd,
 		PeakRateMultiplier:              req.PeakRateMultiplier,
+		ProfitControlEnabled:            req.ProfitControlEnabled,
+		ProfitMinMargin:                 req.ProfitMinMargin,
+		ProfitSafetyBuffer:              req.ProfitSafetyBuffer,
 		ImagePrice1K:                    req.ImagePrice1K,
 		ImagePrice2K:                    req.ImagePrice2K,
 		ImagePrice4K:                    req.ImagePrice4K,
@@ -617,6 +650,7 @@ func (h *GroupHandler) Update(c *gin.Context) {
 		MCPXMLInject:                    req.MCPXMLInject,
 		SupportedModelScopes:            req.SupportedModelScopes,
 		AllowMessagesDispatch:           req.AllowMessagesDispatch,
+		AllowLive:                       req.AllowLive,
 		RequireOAuthOnly:                req.RequireOAuthOnly,
 		RequirePrivacySet:               req.RequirePrivacySet,
 		DefaultMappedModel:              req.DefaultMappedModel,

backend/internal/handler/admin/setting_handler.go

@@ -123,6 +123,7 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 	if paymentCfg == nil {
 		paymentCfg = &service.PaymentConfig{}
 	}
+	passkeyConfigured, passkeyRPID, passkeyRPOrigins := h.settingService.PasskeyConfiguration()
 
 	payload := dto.SystemSettings{
 		RegistrationEnabled:                                    settings.RegistrationEnabled,
@@ -134,6 +135,10 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 		InvitationCodeEnabled:                                  settings.InvitationCodeEnabled,
 		TotpEnabled:                                            settings.TotpEnabled,
 		TotpEncryptionKeyConfigured:                            h.settingService.IsTotpEncryptionKeyConfigured(),
+		PasskeyEnabled:                                         settings.PasskeyEnabled,
+		PasskeyConfigured:                                      passkeyConfigured,
+		PasskeyRPID:                                            passkeyRPID,
+		PasskeyRPOrigins:                                       passkeyRPOrigins,
 		SessionBindingEnabled:                                  settings.SessionBindingEnabled,
 		StepUpEnabled:                                          settings.StepUpEnabled,
 		AuditLogRetentionDays:                                  settings.AuditLogRetentionDays,
@@ -228,6 +233,7 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 		ContactInfo:                                            settings.ContactInfo,
 		DocURL:                                                 settings.DocURL,
 		HomeContent:                                            settings.HomeContent,
+		CompactHomeEnabled:                                     settings.CompactHomeEnabled,
 		HideCcsImportButton:                                    settings.HideCcsImportButton,
 		PurchaseSubscriptionEnabled:                            settings.PurchaseSubscriptionEnabled,
 		PurchaseSubscriptionURL:                                settings.PurchaseSubscriptionURL,
@@ -346,6 +352,10 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 
 		AvailableChannelsEnabled: settings.AvailableChannelsEnabled,
 
+		ModelPlazaEnabled:     settings.ModelPlazaEnabled,
+		ModelPlazaRequireAuth: settings.ModelPlazaRequireAuth,
+		ModelPlazaDescription: settings.ModelPlazaDescription,
+
 		AffiliateEnabled: settings.AffiliateEnabled,
 
 		AllowUserViewErrorRequests: settings.AllowUserViewErrorRequests,

backend/internal/handler/admin/setting_handler_audit.go

@@ -56,6 +56,9 @@ func diffSettings(before *service.SystemSettings, after *service.SystemSettings,
 	if before.TotpEnabled != after.TotpEnabled {
 		changed = append(changed, "totp_enabled")
 	}
+	if before.PasskeyEnabled != after.PasskeyEnabled {
+		changed = append(changed, "passkey_enabled")
+	}
 	if before.SessionBindingEnabled != after.SessionBindingEnabled {
 		changed = append(changed, "session_binding_enabled")
 	}
@@ -296,6 +299,9 @@ func diffSettings(before *service.SystemSettings, after *service.SystemSettings,
 	if before.HomeContent != after.HomeContent {
 		changed = append(changed, "home_content")
 	}
+	if before.CompactHomeEnabled != after.CompactHomeEnabled {
+		changed = append(changed, "compact_home_enabled")
+	}
 	if before.HideCcsImportButton != after.HideCcsImportButton {
 		changed = append(changed, "hide_ccs_import_button")
 	}
@@ -525,6 +531,15 @@ func diffSettings(before *service.SystemSettings, after *service.SystemSettings,
 	if before.AvailableChannelsEnabled != after.AvailableChannelsEnabled {
 		changed = append(changed, "available_channels_enabled")
 	}
+	if before.ModelPlazaEnabled != after.ModelPlazaEnabled {
+		changed = append(changed, "model_plaza_enabled")
+	}
+	if before.ModelPlazaRequireAuth != after.ModelPlazaRequireAuth {
+		changed = append(changed, "model_plaza_require_auth")
+	}
+	if before.ModelPlazaDescription != after.ModelPlazaDescription {
+		changed = append(changed, "model_plaza_description")
+	}
 	if before.AffiliateEnabled != after.AffiliateEnabled {
 		changed = append(changed, "affiliate_enabled")
 	}

backend/internal/handler/admin/setting_handler_partial_payload_test.go

@@ -0,0 +1,67 @@
+//go:build unit
+
+package admin
+
+import (
+	"net/http"
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+
+	"github.com/stretchr/testify/require"
+)
+
+// Saving settings is a whole-document PUT. A client that sends only the field it
+// cares about must not reset everything else: a payload as small as
+// `{"risk_control_enabled":true}` used to clear site_name, after which
+// getStringOrDefault rendered the empty value as the built-in default and the
+// login page silently changed name.
+
+func TestUpdateSettingsPartialPayloadKeepsUnsentKeys(t *testing.T) {
+	h, repo := newStepUpSwitchTestHandler(t, map[string]string{
+		service.SettingKeySiteName:         "Example Gateway",
+		service.SettingKeySiteSubtitle:     "Example Gateway Platform",
+		service.SettingKeySMTPHost:         "smtp.example.com",
+		service.SettingKeySMTPFrom:         "noreply@example.com",
+		service.SettingKeyTurnstileEnabled: "true",
+	})
+
+	rec := doUpdateSettings(t, h, map[string]any{"risk_control_enabled": true}, nil)
+	require.Equal(t, http.StatusOK, rec.Code)
+
+	require.Equal(t, "true", repo.values[service.SettingKeyRiskControlEnabled],
+		"the field the caller actually sent must be written")
+
+	require.Equal(t, "Example Gateway", repo.values[service.SettingKeySiteName])
+	require.Equal(t, "Example Gateway Platform", repo.values[service.SettingKeySiteSubtitle])
+	require.Equal(t, "smtp.example.com", repo.values[service.SettingKeySMTPHost])
+	require.Equal(t, "noreply@example.com", repo.values[service.SettingKeySMTPFrom])
+	require.Equal(t, "true", repo.values[service.SettingKeyTurnstileEnabled])
+}
+
+// A full payload keeps whole-document semantics: fields explicitly set to their
+// zero value are still cleared.
+func TestUpdateSettingsFullPayloadStillClearsSentEmptyFields(t *testing.T) {
+	h, repo := newStepUpSwitchTestHandler(t, map[string]string{
+		service.SettingKeySiteName: "Example Gateway",
+	})
+
+	rec := doUpdateSettings(t, h, map[string]any{"site_name": ""}, nil)
+	require.Equal(t, http.StatusOK, rec.Code)
+
+	require.Equal(t, "", repo.values[service.SettingKeySiteName],
+		"an explicitly sent empty value is a deliberate clear, not an omission")
+}
+
+// smtp_from_email is the one request field whose JSON name differs from its
+// setting key; the alias keeps it from being treated as always-omitted.
+func TestUpdateSettingsSMTPFromAliasIsWritable(t *testing.T) {
+	h, repo := newStepUpSwitchTestHandler(t, map[string]string{
+		service.SettingKeySMTPFrom: "old@example.com",
+	})
+
+	rec := doUpdateSettings(t, h, map[string]any{"smtp_from_email": "new@example.com"}, nil)
+	require.Equal(t, http.StatusOK, rec.Code)
+
+	require.Equal(t, "new@example.com", repo.values[service.SettingKeySMTPFrom])
+}

backend/internal/handler/admin/setting_handler_platform_quota_test.go

@@ -64,6 +64,18 @@ func TestDiffSettings_NoChangeWhenEqual(t *testing.T) {
 	}
 }
 
+func TestDiffSettings_DetectsCompactHomeChange(t *testing.T) {
+	changed := diffSettings(
+		&service.SystemSettings{},
+		&service.SystemSettings{CompactHomeEnabled: true},
+		nil,
+		nil,
+		UpdateSettingsRequest{},
+	)
+
+	require.Contains(t, changed, service.SettingKeyCompactHomeEnabled)
+}
+
 func TestEqualNullableFloat(t *testing.T) {
 	five := 5.0
 	five2 := 5.0

backend/internal/handler/admin/setting_handler_runtime.go

@@ -154,6 +154,70 @@ func (h *SettingHandler) UpdateRateLimit429CooldownSettings(c *gin.Context) {
 	})
 }
 
+// GetPanelRateLimitSettings 获取面板 API 限流配置
+// GET /api/v1/admin/settings/panel-rate-limit
+func (h *SettingHandler) GetPanelRateLimitSettings(c *gin.Context) {
+	settings, err := h.settingService.GetPanelRateLimitSettings(c.Request.Context())
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	response.Success(c, dto.PanelRateLimitSettings{
+		Enabled:     settings.Enabled,
+		UserRPM:     settings.UserRPM,
+		HeavyRPM:    settings.HeavyRPM,
+		ExemptAdmin: settings.ExemptAdmin,
+		PublicIPRPM: settings.PublicIPRPM,
+	})
+}
+
+// UpdatePanelRateLimitSettingsRequest 更新面板 API 限流配置请求
+type UpdatePanelRateLimitSettingsRequest struct {
+	Enabled     bool `json:"enabled"`
+	UserRPM     int  `json:"user_rpm"`
+	HeavyRPM    int  `json:"heavy_rpm"`
+	ExemptAdmin bool `json:"exempt_admin"`
+	PublicIPRPM int  `json:"public_ip_rpm"`
+}
+
+// UpdatePanelRateLimitSettings 更新面板 API 限流配置
+// PUT /api/v1/admin/settings/panel-rate-limit
+func (h *SettingHandler) UpdatePanelRateLimitSettings(c *gin.Context) {
+	var req UpdatePanelRateLimitSettingsRequest
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.BadRequest(c, "Invalid request: "+err.Error())
+		return
+	}
+
+	settings := &service.PanelRateLimitSettings{
+		Enabled:     req.Enabled,
+		UserRPM:     req.UserRPM,
+		HeavyRPM:    req.HeavyRPM,
+		ExemptAdmin: req.ExemptAdmin,
+		PublicIPRPM: req.PublicIPRPM,
+	}
+
+	if err := h.settingService.SetPanelRateLimitSettings(c.Request.Context(), settings); err != nil {
+		response.BadRequest(c, err.Error())
+		return
+	}
+
+	updatedSettings, err := h.settingService.GetPanelRateLimitSettings(c.Request.Context())
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	response.Success(c, dto.PanelRateLimitSettings{
+		Enabled:     updatedSettings.Enabled,
+		UserRPM:     updatedSettings.UserRPM,
+		HeavyRPM:    updatedSettings.HeavyRPM,
+		ExemptAdmin: updatedSettings.ExemptAdmin,
+		PublicIPRPM: updatedSettings.PublicIPRPM,
+	})
+}
+
 // GetStreamTimeoutSettings 获取流超时处理配置
 // GET /api/v1/admin/settings/stream-timeout
 func (h *SettingHandler) GetStreamTimeoutSettings(c *gin.Context) {

backend/internal/handler/admin/setting_handler_update.go

@@ -6,6 +6,7 @@ import (
 	"errors"
 	"log/slog"
 	"net/http"
+	"reflect"
 	"strings"
 
 	"github.com/Wei-Shaw/sub2api/internal/config"
@@ -15,6 +16,7 @@ import (
 	"github.com/Wei-Shaw/sub2api/internal/service"
 
 	"github.com/gin-gonic/gin"
+	"github.com/gin-gonic/gin/binding"
 )
 
 // UpdateSettingsRequest 更新设置请求
@@ -28,6 +30,7 @@ type UpdateSettingsRequest struct {
 	FrontendURL                      string                       `json:"frontend_url"`
 	InvitationCodeEnabled            bool                         `json:"invitation_code_enabled"`
 	TotpEnabled                      bool                         `json:"totp_enabled"`             // TOTP 双因素认证
+	PasskeyEnabled                   *bool                        `json:"passkey_enabled"`          // Passkey 登录（省略=保持现值）
 	SessionBindingEnabled            *bool                        `json:"session_binding_enabled"`  // 会话 IP/UA 绑定（省略=保持现值）
 	StepUpEnabled                    *bool                        `json:"step_up_enabled"`          // 敏感操作 step-up 2FA（省略=保持现值）
 	AuditLogRetentionDays            int                          `json:"audit_log_retention_days"` // 审计日志保留天数
@@ -139,6 +142,7 @@ type UpdateSettingsRequest struct {
 	ContactInfo                 string                `json:"contact_info"`
 	DocURL                      string                `json:"doc_url"`
 	HomeContent                 string                `json:"home_content"`
+	CompactHomeEnabled          bool                  `json:"compact_home_enabled"`
 	HideCcsImportButton         bool                  `json:"hide_ccs_import_button"`
 	PurchaseSubscriptionEnabled *bool                 `json:"purchase_subscription_enabled"`
 	PurchaseSubscriptionURL     *string               `json:"purchase_subscription_url"`
@@ -310,6 +314,11 @@ type UpdateSettingsRequest struct {
 	// Available Channels feature switch (user-facing)
 	AvailableChannelsEnabled *bool `json:"available_channels_enabled"`
 
+	// Model Plaza feature switches + description
+	ModelPlazaEnabled     *bool   `json:"model_plaza_enabled"`
+	ModelPlazaRequireAuth *bool   `json:"model_plaza_require_auth"`
+	ModelPlazaDescription *string `json:"model_plaza_description"`
+
 	// Affiliate (邀请返利) feature switch
 	AffiliateEnabled *bool `json:"affiliate_enabled"`
 
@@ -375,12 +384,70 @@ func (h *SettingHandler) ensureActorTotpForStepUp(c *gin.Context) bool {
 	return true
 }
 
+// settingKeyJSONAliases covers the request fields whose JSON name differs from
+// the setting key they persist to. Every other field of UpdateSettingsRequest
+// is named after its setting key.
+var settingKeyJSONAliases = map[string]string{
+	"smtp_from_email": service.SettingKeySMTPFrom,
+}
+
+// settingKeyByJSONName maps the value-typed top-level JSON fields of
+// UpdateSettingsRequest to the setting key each one writes. Resolved once from
+// the struct tags so new fields are covered without touching this file.
+//
+// Pointer-typed fields are deliberately excluded: they already carry their own
+// "omitted = keep the stored value" merge in UpdateSettings, and some of them
+// rely on being rewritten on every save to re-normalize fail-closed security
+// state (see TestUpdateSettingsMalformedForwardedClientIPHeadersRemainFailClosedWhenOmitted).
+// Only the value-typed fields are indistinguishable from a deliberate clear.
+var settingKeyByJSONName = buildSettingKeyByJSONName()
+
+func buildSettingKeyByJSONName() map[string]string {
+	t := reflect.TypeOf(UpdateSettingsRequest{})
+	out := make(map[string]string, t.NumField())
+	for i := 0; i < t.NumField(); i++ {
+		field := t.Field(i)
+		if field.Type.Kind() == reflect.Ptr {
+			continue
+		}
+		name, _, _ := strings.Cut(field.Tag.Get("json"), ",")
+		if name == "" || name == "-" {
+			continue
+		}
+		if alias, ok := settingKeyJSONAliases[name]; ok {
+			out[name] = alias
+			continue
+		}
+		out[name] = name
+	}
+	return out
+}
+
+// omittedSettingKeys reports the setting keys this payload never mentioned.
+// Saving settings is a whole-document PUT, so without this a client that sends
+// only the one field it cares about resets every other field to a zero value.
+func omittedSettingKeys(sentFields map[string]json.RawMessage) service.OmittedSettingKeys {
+	omitted := make(service.OmittedSettingKeys, len(settingKeyByJSONName))
+	for jsonName, settingKey := range settingKeyByJSONName {
+		if _, sent := sentFields[jsonName]; !sent {
+			omitted[settingKey] = struct{}{}
+		}
+	}
+	return omitted
+}
+
 func (h *SettingHandler) UpdateSettings(c *gin.Context) {
-	var req UpdateSettingsRequest
-	if err := c.ShouldBindJSON(&req); err != nil {
+	var sentFields map[string]json.RawMessage
+	if err := c.ShouldBindBodyWith(&sentFields, binding.JSON); err != nil {
 		response.BadRequest(c, "Invalid request: "+err.Error())
 		return
 	}
+	var req UpdateSettingsRequest
+	if err := c.ShouldBindBodyWith(&req, binding.JSON); err != nil {
+		response.BadRequest(c, "Invalid request: "+err.Error())
+		return
+	}
+	omitted := omittedSettingKeys(sentFields)
 
 	previousSettings, err := h.settingService.GetAllSettings(c.Request.Context())
 	if err != nil {
@@ -403,6 +470,17 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 	if req.StepUpEnabled != nil {
 		stepUpEnabled = *req.StepUpEnabled
 	}
+	passkeyEnabled := previousSettings.PasskeyEnabled
+	if req.PasskeyEnabled != nil {
+		passkeyEnabled = *req.PasskeyEnabled
+	}
+	if passkeyEnabled {
+		configured, _, _ := h.settingService.PasskeyConfiguration()
+		if !configured {
+			response.BadRequest(c, "Passkey sign-in requires a valid WebAuthn RP ID and allowed HTTPS origins in the deployment configuration")
+			return
+		}
+	}
 	forwardedClientIPHeaders := append([]string(nil), previousSettings.ForwardedClientIPHeaders...)
 	if req.ForwardedClientIPHeaders != nil {
 		forwardedClientIPHeaders = append([]string(nil), (*req.ForwardedClientIPHeaders)...)
@@ -1253,6 +1331,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		FrontendURL:                      req.FrontendURL,
 		InvitationCodeEnabled:            req.InvitationCodeEnabled,
 		TotpEnabled:                      req.TotpEnabled,
+		PasskeyEnabled:                   passkeyEnabled,
 		SessionBindingEnabled:            sessionBindingEnabled,
 		StepUpEnabled:                    stepUpEnabled,
 		AuditLogRetentionDays:            req.AuditLogRetentionDays,
@@ -1352,6 +1431,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		ContactInfo:                            req.ContactInfo,
 		DocURL:                                 req.DocURL,
 		HomeContent:                            req.HomeContent,
+		CompactHomeEnabled:                     req.CompactHomeEnabled,
 		HideCcsImportButton:                    req.HideCcsImportButton,
 		PurchaseSubscriptionEnabled:            purchaseEnabled,
 		PurchaseSubscriptionURL:                purchaseURL,
@@ -1605,6 +1685,24 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 			}
 			return previousSettings.AvailableChannelsEnabled
 		}(),
+		ModelPlazaEnabled: func() bool {
+			if req.ModelPlazaEnabled != nil {
+				return *req.ModelPlazaEnabled
+			}
+			return previousSettings.ModelPlazaEnabled
+		}(),
+		ModelPlazaRequireAuth: func() bool {
+			if req.ModelPlazaRequireAuth != nil {
+				return *req.ModelPlazaRequireAuth
+			}
+			return previousSettings.ModelPlazaRequireAuth
+		}(),
+		ModelPlazaDescription: func() string {
+			if req.ModelPlazaDescription != nil {
+				return *req.ModelPlazaDescription
+			}
+			return previousSettings.ModelPlazaDescription
+		}(),
 		AffiliateEnabled: func() bool {
 			if req.AffiliateEnabled != nil {
 				return *req.AffiliateEnabled
@@ -1692,7 +1790,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		},
 		ForceEmailOnThirdPartySignup: boolValueOrDefault(req.ForceEmailOnThirdPartySignup, previousAuthSourceDefaults.ForceEmailOnThirdPartySignup),
 	}
-	if err := h.settingService.UpdateSettingsWithAuthSourceDefaults(c.Request.Context(), settings, authSourceDefaults); err != nil {
+	if err := h.settingService.UpdateSettingsWithAuthSourceDefaultsOmitting(c.Request.Context(), settings, authSourceDefaults, omitted); err != nil {
 		response.ErrorFrom(c, err)
 		return
 	}
@@ -1776,6 +1874,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 	if updatedPaymentCfg == nil {
 		updatedPaymentCfg = &service.PaymentConfig{}
 	}
+	passkeyConfigured, passkeyRPID, passkeyRPOrigins := h.settingService.PasskeyConfiguration()
 
 	payload := dto.SystemSettings{
 		RegistrationEnabled:                                    updatedSettings.RegistrationEnabled,
@@ -1787,6 +1886,10 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		InvitationCodeEnabled:                                  updatedSettings.InvitationCodeEnabled,
 		TotpEnabled:                                            updatedSettings.TotpEnabled,
 		TotpEncryptionKeyConfigured:                            h.settingService.IsTotpEncryptionKeyConfigured(),
+		PasskeyEnabled:                                         updatedSettings.PasskeyEnabled,
+		PasskeyConfigured:                                      passkeyConfigured,
+		PasskeyRPID:                                            passkeyRPID,
+		PasskeyRPOrigins:                                       passkeyRPOrigins,
 		SessionBindingEnabled:                                  updatedSettings.SessionBindingEnabled,
 		StepUpEnabled:                                          updatedSettings.StepUpEnabled,
 		AuditLogRetentionDays:                                  updatedSettings.AuditLogRetentionDays,
@@ -1881,6 +1984,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		ContactInfo:                                            updatedSettings.ContactInfo,
 		DocURL:                                                 updatedSettings.DocURL,
 		HomeContent:                                            updatedSettings.HomeContent,
+		CompactHomeEnabled:                                     updatedSettings.CompactHomeEnabled,
 		HideCcsImportButton:                                    updatedSettings.HideCcsImportButton,
 		PurchaseSubscriptionEnabled:                            updatedSettings.PurchaseSubscriptionEnabled,
 		PurchaseSubscriptionURL:                                updatedSettings.PurchaseSubscriptionURL,
@@ -1995,6 +2099,10 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 
 		AvailableChannelsEnabled: updatedSettings.AvailableChannelsEnabled,
 
+		ModelPlazaEnabled:     updatedSettings.ModelPlazaEnabled,
+		ModelPlazaRequireAuth: updatedSettings.ModelPlazaRequireAuth,
+		ModelPlazaDescription: updatedSettings.ModelPlazaDescription,
+
 		AffiliateEnabled: updatedSettings.AffiliateEnabled,
 
 		RiskControlEnabled:          updatedSettings.RiskControlEnabled,

backend/internal/handler/admin/usage_handler.go

@@ -110,6 +110,7 @@ func (h *UsageHandler) List(c *gin.Context) {
 	}
 
 	model := c.Query("model")
+	requestID := strings.TrimSpace(c.Query("request_id"))
 	billingMode := strings.TrimSpace(c.Query("billing_mode"))
 
 	var requestType *int16
@@ -176,6 +177,7 @@ func (h *UsageHandler) List(c *gin.Context) {
 		APIKeyID:          apiKeyID,
 		AccountID:         accountID,
 		GroupID:           groupID,
+		RequestID:         requestID,
 		Model:             model,
 		ModelFilterSource: usagestats.ModelSourceRequested,
 		RequestType:       requestType,

backend/internal/handler/admin/usage_handler_request_type_test.go

@@ -107,6 +107,18 @@ func TestAdminUsageListExactTotalTrue(t *testing.T) {
 	require.True(t, repo.listFilters.ExactTotal)
 }
 
+func TestAdminUsageListRequestIDFilter(t *testing.T) {
+	repo := &adminUsageRepoCapture{}
+	router := newAdminUsageRequestTypeTestRouter(repo)
+
+	req := httptest.NewRequest(http.MethodGet, "/admin/usage?request_id=req-0123", nil)
+	rec := httptest.NewRecorder()
+	router.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Equal(t, "req-0123", repo.listFilters.RequestID)
+}
+
 func TestAdminUsageListInvalidExactTotal(t *testing.T) {
 	repo := &adminUsageRepoCapture{}
 	router := newAdminUsageRequestTypeTestRouter(repo)

backend/internal/handler/auth_handler.go

@@ -98,16 +98,20 @@ func ensureLoginUserActive(user *service.User) error {
 // respondWithTokenPair 生成 Token 对并返回认证响应
 // 如果 Token 对生成失败，回退到只返回 Access Token（向后兼容）
 func (h *AuthHandler) respondWithTokenPair(c *gin.Context, user *service.User) {
+	respondWithTokenPair(c, h.authService, user)
+}
+
+func respondWithTokenPair(c *gin.Context, authService *service.AuthService, user *service.User) {
 	if err := ensureLoginUserActive(user); err != nil {
 		response.ErrorFrom(c, err)
 		return
 	}
 
-	tokenPair, err := h.authService.GenerateTokenPair(c.Request.Context(), user, "")
+	tokenPair, err := authService.GenerateTokenPair(c.Request.Context(), user, "")
 	if err != nil {
 		slog.Error("failed to generate token pair", "error", err, "user_id", user.ID)
 		// 回退到只返回Access Token
-		token, tokenErr := h.authService.GenerateToken(c.Request.Context(), user)
+		token, tokenErr := authService.GenerateToken(c.Request.Context(), user)
 		if tokenErr != nil {
 			response.InternalError(c, "Failed to generate token")
 			return

backend/internal/handler/auth_oauth_pending_flow_test.go

@@ -2978,6 +2978,17 @@ func (r *oauthPendingFlowUserRepo) Create(ctx context.Context, user *service.Use
 	return nil
 }
 
+func (r *oauthPendingFlowUserRepo) CreateWithEmailAliasGuard(ctx context.Context, user *service.User) error {
+	aliasExists, err := r.ExistsByEmailAlias(ctx, user.Email)
+	if err != nil {
+		return err
+	}
+	if aliasExists {
+		return service.ErrEmailExists
+	}
+	return r.Create(ctx, user)
+}
+
 func (r *oauthPendingFlowUserRepo) GetByID(ctx context.Context, id int64) (*service.User, error) {
 	entity, err := r.client.User.Get(ctx, id)
 	if err != nil {
@@ -3004,7 +3015,7 @@ func (r *oauthPendingFlowUserRepo) GetFirstAdmin(context.Context) (*service.User
 	panic("unexpected GetFirstAdmin call")
 }
 
-func (r *oauthPendingFlowUserRepo) Update(ctx context.Context, user *service.User) error {
+func (r *oauthPendingFlowUserRepo) Update(ctx context.Context, user *service.User, fields service.UserUpdateFields) error {
 	entity, err := r.client.User.UpdateOneID(user.ID).
 		SetEmail(user.Email).
 		SetUsername(user.Username).
@@ -3157,6 +3168,14 @@ func (r *oauthPendingFlowUserRepo) DeductBalance(context.Context, int64, float64
 	panic("unexpected DeductBalance call")
 }
 
+func (r *oauthPendingFlowUserRepo) AdjustBalance(ctx context.Context, id int64, delta float64) (service.BalanceChange, error) {
+	panic("unexpected AdjustBalance call")
+}
+
+func (r *oauthPendingFlowUserRepo) SetBalance(ctx context.Context, id int64, value float64) (service.BalanceChange, error) {
+	panic("unexpected SetBalance call")
+}
+
 func (r *oauthPendingFlowUserRepo) UpdateConcurrency(context.Context, int64, int) error {
 	panic("unexpected UpdateConcurrency call")
 }
@@ -3185,6 +3204,20 @@ func (r *oauthPendingFlowUserRepo) ExistsByEmail(ctx context.Context, email stri
 	return count > 0, err
 }
 
+func (r *oauthPendingFlowUserRepo) ExistsByEmailAlias(ctx context.Context, email string) (bool, error) {
+	identity := service.NormalizeEmailForAliasDedup(email)
+	emails, err := r.client.User.Query().Select(dbuser.FieldEmail).Strings(ctx)
+	if err != nil {
+		return false, err
+	}
+	for _, stored := range emails {
+		if service.NormalizeEmailForAliasDedup(stored) == identity {
+			return true, nil
+		}
+	}
+	return false, nil
+}
+
 func (r *oauthPendingFlowUserRepo) RemoveGroupFromAllowedGroups(context.Context, int64) (int64, error) {
 	panic("unexpected RemoveGroupFromAllowedGroups call")
 }

backend/internal/handler/auth_session_revocation_test.go

@@ -47,7 +47,11 @@ func TestAuthHandlerRevokeAllSessionsInvalidatesAccessTokens(t *testing.T) {
 
 	require.Equal(t, http.StatusOK, recorder.Code)
 	require.Equal(t, []int64{29}, refreshTokenCache.revokedUserIDs)
-	require.Equal(t, int64(8), repo.user.TokenVersion)
+	// users 表没有 token_version 列（见 resolvedTokenVersion：JWT 里的值由
+	// email+password_hash 指纹推导），所以自增 TokenVersion 只停留在内存里。
+	// 此前紧跟其后的整行 Update 不写任何有效数据，却会用旧快照覆盖并发写入的列，
+	// 已移除。会话撤销由上面的 refresh session 清理承担。
+	require.Equal(t, int64(7), repo.user.TokenVersion)
 
 	var resp struct {
 		Code int `json:"code"`

backend/internal/handler/available_channel_handler.go

@@ -12,12 +12,11 @@ import (
 
 // AvailableChannelHandler 处理用户侧「可用渠道」查询。
 //
-// 用户侧接口委托 ChannelService.ListAvailable，并在返回前做三层过滤：
+// 用户侧接口委托 ChannelService.ListAvailable，并在返回前做四层过滤：
 //  1. 行过滤：只保留状态为 Active 且与当前用户可访问分组有交集的渠道；
 //  2. 分组过滤：渠道的 Groups 只保留用户可访问的那些；
-//  3. 平台过滤：渠道的 SupportedModels 只保留平台在用户可见 Groups 中出现过的模型，
-//     防止"渠道同时挂在 antigravity / anthropic 两个平台的分组上，用户只访问
-//     antigravity，却看到 anthropic 模型"这类跨平台信息泄漏；
+//  3. 平台过滤：普通分组只保留自身平台模型；Composite 分组按渠道已配置的具体模型平台
+//     展开。这样既防止普通分组跨平台泄漏，也让 Composite 正确展示其多平台能力；
 //  4. 字段白名单：仅返回用户需要的字段（省略 BillingModelSource / RestrictModels
 //     / 内部 ID / Status 等管理字段）。
 type AvailableChannelHandler struct {
@@ -172,19 +171,47 @@ func (h *AvailableChannelHandler) List(c *gin.Context) {
 }
 
 // buildPlatformSections 把一个渠道按 visibleGroups 的平台集合拆成有序的 section 列表：
-// 每个 section 对应一个平台，只包含该平台的 groups 和 supported_models。
+// 每个 section 对应一个具体平台，只包含该平台的 groups 和 supported_models。
+//
+// Composite 分组可访问渠道中所有已配置的具体平台，因此会被展开到每个有支持模型的
+// 平台 section。普通分组仍严格留在自身平台，避免跨平台模型信息泄漏。Composite 渠道
+// 尚未配置任何模型时保留 composite section，以便前端继续展示该分组和“未配置模型”状态。
 // 输出按 platform 字母序稳定排序，便于前端等效比较与回归测试。
 func buildPlatformSections(
 	ch service.AvailableChannel,
 	visibleGroups []userAvailableGroup,
 ) []userChannelPlatformSection {
 	groupsByPlatform := make(map[string][]userAvailableGroup, 4)
+	compositeGroups := make([]userAvailableGroup, 0, 1)
 	for _, g := range visibleGroups {
 		if g.Platform == "" {
 			continue
 		}
+		if g.Platform == service.PlatformComposite {
+			compositeGroups = append(compositeGroups, g)
+			continue
+		}
 		groupsByPlatform[g.Platform] = append(groupsByPlatform[g.Platform], g)
 	}
+
+	if len(compositeGroups) > 0 {
+		modelPlatforms := make(map[string]struct{}, len(ch.SupportedModels))
+		for i := range ch.SupportedModels {
+			if platform := ch.SupportedModels[i].Platform; platform != "" {
+				modelPlatforms[platform] = struct{}{}
+			}
+		}
+		if len(modelPlatforms) == 0 {
+			groupsByPlatform[service.PlatformComposite] = append(
+				groupsByPlatform[service.PlatformComposite],
+				compositeGroups...,
+			)
+		} else {
+			for platform := range modelPlatforms {
+				groupsByPlatform[platform] = append(groupsByPlatform[platform], compositeGroups...)
+			}
+		}
+	}
 	if len(groupsByPlatform) == 0 {
 		return nil
 	}

backend/internal/handler/available_channel_handler_test.go

@@ -155,3 +155,101 @@ func TestBuildPlatformSections_GroupsByPlatform(t *testing.T) {
 	require.Len(t, sections[0].SupportedModels, 1)
 	require.Equal(t, "claude-sonnet-4-6", sections[0].SupportedModels[0].Name)
 }
+
+func TestBuildPlatformSections_CompositeGroupExpandsAcrossConfiguredModelPlatforms(t *testing.T) {
+	anthropicPrice := 3e-6
+	openAIPrice := 2.5e-6
+	ch := service.AvailableChannel{
+		Name: "composite-channel",
+		SupportedModels: []service.SupportedModel{
+			{
+				Name:     "claude-sonnet-4-6",
+				Platform: service.PlatformAnthropic,
+				Pricing:  &service.ChannelModelPricing{InputPrice: &anthropicPrice},
+			},
+			{
+				Name:     "gpt-5",
+				Platform: service.PlatformOpenAI,
+				Pricing:  &service.ChannelModelPricing{InputPrice: &openAIPrice},
+			},
+		},
+	}
+	visible := []userAvailableGroup{
+		{ID: 9, Name: "composite", Platform: service.PlatformComposite},
+	}
+
+	sections := buildPlatformSections(ch, visible)
+
+	require.Len(t, sections, 2)
+	require.Equal(t, service.PlatformAnthropic, sections[0].Platform)
+	require.Equal(t, service.PlatformOpenAI, sections[1].Platform)
+	for _, section := range sections {
+		require.Len(t, section.Groups, 1)
+		require.Equal(t, int64(9), section.Groups[0].ID)
+		require.Equal(t, service.PlatformComposite, section.Groups[0].Platform)
+		require.Len(t, section.SupportedModels, 1)
+		require.Equal(t, section.Platform, section.SupportedModels[0].Platform)
+		require.NotNil(t, section.SupportedModels[0].Pricing)
+	}
+	require.Equal(t, "claude-sonnet-4-6", sections[0].SupportedModels[0].Name)
+	require.Equal(t, "gpt-5", sections[1].SupportedModels[0].Name)
+}
+
+func TestBuildPlatformSections_OrdinaryGroupRemainsPlatformIsolated(t *testing.T) {
+	ch := service.AvailableChannel{
+		SupportedModels: []service.SupportedModel{
+			{Name: "claude-sonnet-4-6", Platform: service.PlatformAnthropic},
+			{Name: "gpt-5", Platform: service.PlatformOpenAI},
+		},
+	}
+	visible := []userAvailableGroup{
+		{ID: 1, Name: "anthropic-only", Platform: service.PlatformAnthropic},
+	}
+
+	sections := buildPlatformSections(ch, visible)
+
+	require.Len(t, sections, 1)
+	require.Equal(t, service.PlatformAnthropic, sections[0].Platform)
+	require.Len(t, sections[0].SupportedModels, 1)
+	require.Equal(t, "claude-sonnet-4-6", sections[0].SupportedModels[0].Name)
+}
+
+func TestBuildPlatformSections_CompositeAndOrdinaryGroupsShareConcreteSection(t *testing.T) {
+	ch := service.AvailableChannel{
+		SupportedModels: []service.SupportedModel{
+			{Name: "claude-sonnet-4-6", Platform: service.PlatformAnthropic},
+			{Name: "gpt-5", Platform: service.PlatformOpenAI},
+		},
+	}
+	visible := []userAvailableGroup{
+		{ID: 1, Name: "anthropic-only", Platform: service.PlatformAnthropic},
+		{ID: 9, Name: "composite", Platform: service.PlatformComposite},
+	}
+
+	sections := buildPlatformSections(ch, visible)
+
+	require.Len(t, sections, 2)
+	require.Equal(t, service.PlatformAnthropic, sections[0].Platform)
+	require.Equal(t, []int64{1, 9}, []int64{
+		sections[0].Groups[0].ID,
+		sections[0].Groups[1].ID,
+	})
+	require.Equal(t, service.PlatformOpenAI, sections[1].Platform)
+	require.Len(t, sections[1].Groups, 1)
+	require.Equal(t, int64(9), sections[1].Groups[0].ID)
+}
+
+func TestBuildPlatformSections_CompositeWithoutModelsKeepsEmptyCompositeSection(t *testing.T) {
+	visible := []userAvailableGroup{
+		{ID: 9, Name: "composite", Platform: service.PlatformComposite},
+	}
+
+	sections := buildPlatformSections(service.AvailableChannel{
+		SupportedModels: []service.SupportedModel{{Name: "invalid-without-platform"}},
+	}, visible)
+
+	require.Len(t, sections, 1)
+	require.Equal(t, service.PlatformComposite, sections[0].Platform)
+	require.Len(t, sections[0].Groups, 1)
+	require.Empty(t, sections[0].SupportedModels)
+}

backend/internal/handler/batch_image_handler.go

@@ -42,6 +42,9 @@ func (h *BatchImageHandler) Submit(c *gin.Context) {
 	if !h.checkSecurityAuditBeforeSubmit(c, &req) {
 		return
 	}
+	if sessionID := service.ExtractClientSessionID(c); sessionID != "" {
+		req.SessionID = &sessionID
+	}
 	got, err := h.service.Submit(c.Request.Context(), owner, req, c.GetHeader("Idempotency-Key"))
 	if err != nil {
 		batchImageError(c, err)

backend/internal/handler/dto/group_mapper_profit_visibility_test.go

@@ -0,0 +1,75 @@
+package dto
+
+import (
+	"encoding/json"
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+)
+
+// profitControlJSONFields 是分组利润控制的三个 JSON 字段。它们与同响应中的
+// rate_multiplier 相乘即可反推出运营方的上游采购成本上限，属于内部经营信息，
+// 只能出现在管理员 DTO 中。
+var profitControlJSONFields = []string{
+	"profit_control_enabled",
+	"profit_min_margin",
+	"profit_safety_buffer",
+}
+
+func profitControlServiceGroup() *service.Group {
+	return &service.Group{
+		ID:                   7,
+		Name:                 "profit-gated",
+		Platform:             service.PlatformAnthropic,
+		RateMultiplier:       2.0,
+		Status:               service.StatusActive,
+		ProfitControlEnabled: true,
+		ProfitMinMargin:      0.3,
+		ProfitSafetyBuffer:   0.05,
+	}
+}
+
+func marshalToMap(t *testing.T, v any) map[string]any {
+	t.Helper()
+	raw, err := json.Marshal(v)
+	if err != nil {
+		t.Fatalf("marshal: %v", err)
+	}
+	var out map[string]any
+	if err := json.Unmarshal(raw, &out); err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+	return out
+}
+
+// TestGroupFromServiceOmitsProfitControl 钉死普通用户侧的分组 DTO 不泄露利润控制配置。
+func TestGroupFromServiceOmitsProfitControl(t *testing.T) {
+	for name, got := range map[string]any{
+		"GroupFromService":        GroupFromService(profitControlServiceGroup()),
+		"GroupFromServiceShallow": GroupFromServiceShallow(profitControlServiceGroup()),
+	} {
+		fields := marshalToMap(t, got)
+		for _, f := range profitControlJSONFields {
+			if _, ok := fields[f]; ok {
+				t.Errorf("%s: 普通用户 DTO 不得包含 %q", name, f)
+			}
+		}
+		if _, ok := fields["rate_multiplier"]; !ok {
+			t.Errorf("%s: 应仍返回 rate_multiplier", name)
+		}
+	}
+}
+
+// TestGroupFromServiceAdminIncludesProfitControl 钉死管理端仍能读写利润控制配置。
+func TestGroupFromServiceAdminIncludesProfitControl(t *testing.T) {
+	admin := GroupFromServiceAdmin(profitControlServiceGroup())
+	if admin.ProfitControlEnabled != true || admin.ProfitMinMargin != 0.3 || admin.ProfitSafetyBuffer != 0.05 {
+		t.Fatalf("管理员 DTO 未透传利润控制配置: %+v", admin)
+	}
+	fields := marshalToMap(t, admin)
+	for _, f := range profitControlJSONFields {
+		if _, ok := fields[f]; !ok {
+			t.Errorf("管理员 DTO 应包含 %q", f)
+		}
+	}
+}

backend/internal/handler/dto/mappers.go

@@ -146,6 +146,9 @@ func GroupFromServiceAdmin(g *service.Group) *AdminGroup {
 	}
 	out := &AdminGroup{
 		Group:                       groupFromServiceBase(g),
+		ProfitControlEnabled:        g.ProfitControlEnabled,
+		ProfitMinMargin:             g.ProfitMinMargin,
+		ProfitSafetyBuffer:          g.ProfitSafetyBuffer,
 		ModelRouting:                g.ModelRouting,
 		ModelRoutingEnabled:         g.ModelRoutingEnabled,
 		MCPXMLInject:                g.MCPXMLInject,
@@ -204,6 +207,7 @@ func groupFromServiceBase(g *service.Group) Group {
 		FallbackGroupID:                 g.FallbackGroupID,
 		FallbackGroupIDOnInvalidRequest: g.FallbackGroupIDOnInvalidRequest,
 		AllowMessagesDispatch:           g.AllowMessagesDispatch,
+		AllowLive:                       g.AllowLive,
 		RequireOAuthOnly:                g.RequireOAuthOnly,
 		RequirePrivacySet:               g.RequirePrivacySet,
 		RPMLimit:                        g.RPMLimit,
@@ -669,6 +673,7 @@ func usageLogFromServiceUser(l *service.UsageLog) UsageLog {
 		MediaType:                 l.MediaType,
 		UserAgent:                 l.UserAgent,
 		IPAddress:                 l.IPAddress,
+		SessionID:                 l.SessionID,
 		CacheTTLOverridden:        l.CacheTTLOverridden,
 		BillingMode:               l.BillingMode,
 		CreatedAt:                 l.CreatedAt,

backend/internal/handler/dto/settings.go

@@ -36,9 +36,13 @@ type SystemSettings struct {
 	InvitationCodeEnabled            bool                     `json:"invitation_code_enabled"`
 	TotpEnabled                      bool                     `json:"totp_enabled"`                   // TOTP 双因素认证
 	TotpEncryptionKeyConfigured      bool                     `json:"totp_encryption_key_configured"` // TOTP 加密密钥是否已配置
-	SessionBindingEnabled            bool                     `json:"session_binding_enabled"`        // 会话 IP/UA 绑定
-	StepUpEnabled                    bool                     `json:"step_up_enabled"`                // 敏感操作 step-up 2FA
-	AuditLogRetentionDays            int                      `json:"audit_log_retention_days"`       // 审计日志保留天数
+	PasskeyEnabled                   bool                     `json:"passkey_enabled"`
+	PasskeyConfigured                bool                     `json:"passkey_configured"`
+	PasskeyRPID                      string                   `json:"passkey_rp_id"`
+	PasskeyRPOrigins                 []string                 `json:"passkey_rp_origins"`
+	SessionBindingEnabled            bool                     `json:"session_binding_enabled"`  // 会话 IP/UA 绑定
+	StepUpEnabled                    bool                     `json:"step_up_enabled"`          // 敏感操作 step-up 2FA
+	AuditLogRetentionDays            int                      `json:"audit_log_retention_days"` // 审计日志保留天数
 	LoginAgreementEnabled            bool                     `json:"login_agreement_enabled"`
 	LoginAgreementMode               string                   `json:"login_agreement_mode"`
 	LoginAgreementUpdatedAt          string                   `json:"login_agreement_updated_at"`
@@ -138,6 +142,7 @@ type SystemSettings struct {
 	ContactInfo                 string           `json:"contact_info"`
 	DocURL                      string           `json:"doc_url"`
 	HomeContent                 string           `json:"home_content"`
+	CompactHomeEnabled          bool             `json:"compact_home_enabled"`
 	HideCcsImportButton         bool             `json:"hide_ccs_import_button"`
 	PurchaseSubscriptionEnabled bool             `json:"purchase_subscription_enabled"`
 	PurchaseSubscriptionURL     string           `json:"purchase_subscription_url"`
@@ -286,6 +291,11 @@ type SystemSettings struct {
 	// Available Channels feature switch (user-facing aggregate view)
 	AvailableChannelsEnabled bool `json:"available_channels_enabled"`
 
+	// Model Plaza feature (public group/model pricing showcase)
+	ModelPlazaEnabled     bool   `json:"model_plaza_enabled"`
+	ModelPlazaRequireAuth bool   `json:"model_plaza_require_auth"`
+	ModelPlazaDescription string `json:"model_plaza_description"`
+
 	// 风控中心功能开关
 	RiskControlEnabled bool `json:"risk_control_enabled"`
 
@@ -320,6 +330,7 @@ type PublicSettings struct {
 	PasswordResetEnabled             bool                     `json:"password_reset_enabled"`
 	InvitationCodeEnabled            bool                     `json:"invitation_code_enabled"`
 	TotpEnabled                      bool                     `json:"totp_enabled"` // TOTP 双因素认证
+	PasskeyEnabled                   bool                     `json:"passkey_enabled"`
 	LoginAgreementEnabled            bool                     `json:"login_agreement_enabled"`
 	LoginAgreementMode               string                   `json:"login_agreement_mode"`
 	LoginAgreementUpdatedAt          string                   `json:"login_agreement_updated_at"`
@@ -334,6 +345,7 @@ type PublicSettings struct {
 	ContactInfo                      string                   `json:"contact_info"`
 	DocURL                           string                   `json:"doc_url"`
 	HomeContent                      string                   `json:"home_content"`
+	CompactHomeEnabled               bool                     `json:"compact_home_enabled"`
 	HideCcsImportButton              bool                     `json:"hide_ccs_import_button"`
 	PurchaseSubscriptionEnabled      bool                     `json:"purchase_subscription_enabled"`
 	PurchaseSubscriptionURL          string                   `json:"purchase_subscription_url"`
@@ -369,6 +381,9 @@ type PublicSettings struct {
 
 	AvailableChannelsEnabled bool `json:"available_channels_enabled"`
 
+	ModelPlazaEnabled     bool `json:"model_plaza_enabled"`
+	ModelPlazaRequireAuth bool `json:"model_plaza_require_auth"`
+
 	AffiliateEnabled bool `json:"affiliate_enabled"`
 
 	RiskControlEnabled bool `json:"risk_control_enabled"`
@@ -394,6 +409,15 @@ type RateLimit429CooldownSettings struct {
 	CooldownSeconds int  `json:"cooldown_seconds"`
 }
 
+// PanelRateLimitSettings 面板 API 限流配置 DTO
+type PanelRateLimitSettings struct {
+	Enabled     bool `json:"enabled"`
+	UserRPM     int  `json:"user_rpm"`
+	HeavyRPM    int  `json:"heavy_rpm"`
+	ExemptAdmin bool `json:"exempt_admin"`
+	PublicIPRPM int  `json:"public_ip_rpm"`
+}
+
 // StreamTimeoutSettings 流超时处理配置 DTO
 type StreamTimeoutSettings struct {
 	Enabled                bool   `json:"enabled"`

backend/internal/handler/dto/types.go

@@ -132,6 +132,8 @@ type Group struct {
 
 	// OpenAI Messages 调度开关（用户侧需要此字段判断是否展示 Claude Code 教程）
 	AllowMessagesDispatch bool `json:"allow_messages_dispatch"`
+	// OpenAI Live 接口开关
+	AllowLive bool `json:"allow_live"`
 
 	// 账号过滤控制（仅 OpenAI/Antigravity 平台有效）
 	RequireOAuthOnly  bool `json:"require_oauth_only"`
@@ -153,6 +155,13 @@ type Group struct {
 type AdminGroup struct {
 	Group
 
+	// 分组利润控制（五个 token 平台分组可启用；margin/buffer 为小数存储）。
+	// 仅管理员可见：这三个字段与同响应中的 rate_multiplier 相乘即可反推出
+	// 运营方的上游成本上限，属于内部经营信息，不得下放到 dto.Group。
+	ProfitControlEnabled bool    `json:"profit_control_enabled"`
+	ProfitMinMargin      float64 `json:"profit_min_margin"`
+	ProfitSafetyBuffer   float64 `json:"profit_safety_buffer"`
+
 	// 模型路由配置（仅 anthropic 平台使用）
 	ModelRouting        map[string][]int64 `json:"model_routing"`
 	ModelRoutingEnabled bool               `json:"model_routing_enabled"`
@@ -521,6 +530,9 @@ type UsageLog struct {
 	UserAgent *string `json:"user_agent"`
 	// IPAddress is visible to the owner of the usage record.
 	IPAddress *string `json:"ip_address,omitempty"`
+	// SessionID is the explicit client-provided request correlation identifier
+	// (e.g. the session_id / X-Session-Id headers). Omitted when absent.
+	SessionID *string `json:"session_id,omitempty"`
 
 	// Cache TTL Override 标记
 	CacheTTLOverridden bool `json:"cache_ttl_overridden"`

backend/internal/handler/endpoint.go

@@ -31,9 +31,12 @@ const (
 	EndpointGeminiModels      = "/v1beta/models"
 )
 
+const EndpointAntigravityGenerateContent = "/v1internal:streamGenerateContent"
+
 // gin.Context keys used by the middleware and helpers below.
 const (
-	ctxKeyInboundEndpoint = "_gateway_inbound_endpoint"
+	ctxKeyInboundEndpoint        = "_gateway_inbound_endpoint"
+	ctxKeyActualUpstreamEndpoint = "_gateway_actual_upstream_endpoint"
 )
 
 // ──────────────────────────────────────────────────────────
@@ -297,6 +300,13 @@ func GetInboundEndpoint(c *gin.Context) string {
 // and the account platform. Handlers call this after scheduling an
 // account, passing account.Platform.
 func GetUpstreamEndpoint(c *gin.Context, platform string) string {
+	if c != nil {
+		if value, ok := c.Get(ctxKeyActualUpstreamEndpoint); ok {
+			if endpoint, ok := value.(string); ok && endpoint != "" {
+				return endpoint
+			}
+		}
+	}
 	inbound := GetInboundEndpoint(c)
 	rawPath := ""
 	if c != nil && c.Request != nil && c.Request.URL != nil {
@@ -304,3 +314,15 @@ func GetUpstreamEndpoint(c *gin.Context, platform string) string {
 	}
 	return DeriveUpstreamEndpoint(inbound, rawPath, platform)
 }
+
+func setActualUpstreamEndpoint(c *gin.Context, endpoint string) {
+	if c != nil {
+		c.Set(ctxKeyActualUpstreamEndpoint, strings.TrimSpace(endpoint))
+	}
+}
+
+func shouldUseAntigravityCompat(account *service.Account) bool {
+	return account != nil &&
+		account.Platform == service.PlatformAntigravity &&
+		account.Type == service.AccountTypeOAuth
+}

backend/internal/handler/endpoint_test.go

@@ -146,6 +146,40 @@ func TestDeriveUpstreamEndpoint(t *testing.T) {
 	}
 }
 
+func TestShouldUseAntigravityCompat(t *testing.T) {
+	tests := []struct {
+		name    string
+		account *service.Account
+		want    bool
+	}{
+		{"oauth", &service.Account{Platform: service.PlatformAntigravity, Type: service.AccountTypeOAuth}, true},
+		{"setup token", &service.Account{Platform: service.PlatformAntigravity, Type: service.AccountTypeSetupToken}, false},
+		{"upstream", &service.Account{Platform: service.PlatformAntigravity, Type: service.AccountTypeUpstream}, false},
+		{"api key", &service.Account{Platform: service.PlatformAntigravity, Type: service.AccountTypeAPIKey}, false},
+		{"anthropic oauth", &service.Account{Platform: service.PlatformAnthropic, Type: service.AccountTypeOAuth}, false},
+		{"nil", nil, false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			require.Equal(t, tt.want, shouldUseAntigravityCompat(tt.account))
+		})
+	}
+}
+
+func TestGetUpstreamEndpointPrefersRuntimeOverride(t *testing.T) {
+	recorder := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(recorder)
+	c.Request = httptest.NewRequest(http.MethodPost, EndpointChatCompletions, nil)
+	c.Set(ctxKeyInboundEndpoint, EndpointChatCompletions)
+
+	setActualUpstreamEndpoint(c, EndpointAntigravityGenerateContent)
+	require.Equal(t, EndpointAntigravityGenerateContent, GetUpstreamEndpoint(c, service.PlatformAntigravity))
+
+	setActualUpstreamEndpoint(c, "")
+	require.Equal(t, EndpointMessages, GetUpstreamEndpoint(c, service.PlatformAntigravity))
+}
+
 func TestResolveOpenAIUpstreamEndpointPrefersForwardResult(t *testing.T) {
 	tests := []struct {
 		name            string

backend/internal/handler/failover_loop.go

@@ -40,8 +40,18 @@ const (
 	// Service 层在 SingleAccountRetry 模式下已做充分原地重试（最多 3 次、总等待 30s），
 	// Handler 层只需短暂间隔后重新进入 Service 层即可。
 	singleAccountBackoffDelay = 2 * time.Second
+	// maxProfitVetoAttempts 单次请求内允许的分组利润门终检否决次数上限。
+	// 利润否决不产生上游请求，因此不会推进 SwitchCount；没有独立上限的话，
+	// 「选号 → 终检否决 → 重选」在候选池与账号快照短暂不一致时可以空转很久。
+	// 取值与 maxAccountSwitches 默认值一致：混合定价的大分组仍有充分重选机会，
+	// 同时把整池越线时的无谓选号开销限制在常数级。
+	maxProfitVetoAttempts = 10
 )
 
+// profitVetoExhaustedMessage 是利润否决次数耗尽时返回给客户端的文案。
+// 语义上等同于「无可用账号」：候选账号都不满足分组的利润约束。
+const profitVetoExhaustedMessage = "No available accounts: all candidates rejected by group profit control"
+
 // FailoverState 跨循环迭代共享的 failover 状态
 type FailoverState struct {
 	SwitchCount           int
@@ -51,18 +61,64 @@ type FailoverState struct {
 	LastFailoverErr       *service.UpstreamFailoverError
 	ForceCacheBilling     bool
 	hasBoundSession       bool
+
+	// profitVetoedAccountIDs 记录被分组利润门终检否决的账号，是 FailedAccountIDs
+	// 的子集。之所以单独维护：HandleSelectionExhausted 的 503 退避分支会清空
+	// FailedAccountIDs，而利润否决在同一请求内的判定不会改变（下游倍率 D 已在
+	// 请求开始冻结），被清空的账号会被立即重选并再次否决，形成没有任何上游请求、
+	// SwitchCount 也不前进的活锁。清空后必须把它们放回排除集。
+	profitVetoedAccountIDs map[int64]struct{}
+	// profitVetoCount 本次请求累计的利润否决次数，用于 maxProfitVetoAttempts 上限。
+	profitVetoCount int
 }
 
 // NewFailoverState 创建 failover 状态
 func NewFailoverState(maxSwitches int, hasBoundSession bool) *FailoverState {
 	return &FailoverState{
-		MaxSwitches:           maxSwitches,
-		FailedAccountIDs:      make(map[int64]struct{}),
-		SameAccountRetryCount: make(map[int64]int),
-		hasBoundSession:       hasBoundSession,
+		MaxSwitches:            maxSwitches,
+		FailedAccountIDs:       make(map[int64]struct{}),
+		SameAccountRetryCount:  make(map[int64]int),
+		hasBoundSession:        hasBoundSession,
+		profitVetoedAccountIDs: make(map[int64]struct{}),
 	}
 }
 
+// RecordProfitVeto 记录一次分组利润门终检否决：把账号加入排除列表（同时登记到
+// 利润否决集，使其不被 503 退避分支清掉）并递增否决计数。
+//
+// 返回 FailoverContinue 表示调用方可以继续重选下一个账号；返回 FailoverExhausted
+// 表示本次请求的利润否决次数已达上限，调用方应按「无可用账号」终止，
+// 不得继续 continue。
+func (s *FailoverState) RecordProfitVeto(accountID int64) FailoverAction {
+	s.FailedAccountIDs[accountID] = struct{}{}
+	if s.profitVetoedAccountIDs == nil {
+		s.profitVetoedAccountIDs = make(map[int64]struct{})
+	}
+	s.profitVetoedAccountIDs[accountID] = struct{}{}
+	s.profitVetoCount++
+	if s.profitVetoCount >= maxProfitVetoAttempts {
+		return FailoverExhausted
+	}
+	return FailoverContinue
+}
+
+// ProfitVetoCount 返回本次请求累计的利润否决次数（供日志使用）。
+func (s *FailoverState) ProfitVetoCount() int { return s.profitVetoCount }
+
+// allExclusionsAreProfitVetoed 判断排除列表是否已全部由利润门否决贡献。
+// 此时清空 FailedAccountIDs 会被原样恢复，退避重试不会带来任何新候选。
+func (s *FailoverState) allExclusionsAreProfitVetoed() bool {
+	if len(s.profitVetoedAccountIDs) == 0 || len(s.FailedAccountIDs) == 0 {
+		return false
+	}
+	for id := range s.FailedAccountIDs {
+		if _, ok := s.profitVetoedAccountIDs[id]; !ok {
+			return false
+		}
+	}
+	return true
+}
+
 // HandleFailoverError 处理 UpstreamFailoverError，返回下一步动作。
 // 包含：缓存计费判断、同账号重试、临时封禁、切换计数、Antigravity 延时。
 func (s *FailoverState) HandleFailoverError(
@@ -156,6 +212,17 @@ func (s *FailoverState) HandleSelectionExhausted(ctx context.Context) FailoverAc
 		s.LastFailoverErr.StatusCode == http.StatusServiceUnavailable &&
 		s.SwitchCount <= s.MaxSwitches {
 
+		// 排除列表全由利润门否决贡献时，清空后会被原样恢复：退避重试拿不到
+		// 任何新候选，而利润否决不推进 SwitchCount，退避条件将永远成立。
+		// 这里直接判定耗尽，避免每 2s 空转一轮的活锁。
+		if s.allExclusionsAreProfitVetoed() {
+			logger.FromContext(ctx).Warn("gateway.failover_selection_exhausted_by_profit_veto",
+				zap.Int("profit_veto_count", s.profitVetoCount),
+				zap.Int("excluded_accounts", len(s.FailedAccountIDs)),
+			)
+			return FailoverExhausted
+		}
+
 		logger.FromContext(ctx).Warn("gateway.failover_single_account_backoff",
 			zap.Duration("backoff_delay", singleAccountBackoffDelay),
 			zap.Int("switch_count", s.SwitchCount),
@@ -169,6 +236,11 @@ func (s *FailoverState) HandleSelectionExhausted(ctx context.Context) FailoverAc
 			zap.Int("max_switches", s.MaxSwitches),
 		)
 		s.FailedAccountIDs = make(map[int64]struct{})
+		// 利润门否决的账号不参与退避重试的解除：判定依据（冻结的下游倍率）在
+		// 同一请求内不变，放它们回池只会被再次否决。
+		for id := range s.profitVetoedAccountIDs {
+			s.FailedAccountIDs[id] = struct{}{}
+		}
 		return FailoverContinue
 	}
 	return FailoverExhausted

backend/internal/handler/failover_loop_profit_veto_test.go

@@ -0,0 +1,139 @@
+package handler
+
+import (
+	"context"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+)
+
+// profitVetoLoopResult 记录一次模拟选号循环的终止方式与步数。
+type profitVetoLoopResult struct {
+	outcome        string // "forwarded" | "exhausted" | "budget_exceeded"
+	forwardedID    int64
+	iterations     int
+	backoffRetries int
+}
+
+// runProfitVetoLoop 模拟 handler 的「选号 → 利润终检 → 重选」循环，只保留与
+// 活锁相关的状态机（FailoverState + 排除列表 + HandleSelectionExhausted），
+// 不涉及真实调度器与上游转发。
+//
+// pool 按顺序给出候选账号；vetoed 中的账号每次终检都被利润门否决——这对应
+// 「候选池快照与 per-account 快照短暂不一致」时门的确定性判定。
+// maxIterations 是测试自身的预算：循环超过它即视为活锁。
+func runProfitVetoLoop(t *testing.T, fs *FailoverState, pool []int64, vetoed map[int64]bool, maxIterations int) profitVetoLoopResult {
+	t.Helper()
+	res := profitVetoLoopResult{}
+	for res.iterations = 1; res.iterations <= maxIterations; res.iterations++ {
+		// 选号：返回第一个不在排除列表中的账号。
+		var picked int64
+		for _, id := range pool {
+			if _, excluded := fs.FailedAccountIDs[id]; !excluded {
+				picked = id
+				break
+			}
+		}
+		if picked == 0 {
+			// 选号耗尽，交给退避决策。
+			switch fs.HandleSelectionExhausted(context.Background()) {
+			case FailoverContinue:
+				res.backoffRetries++
+				continue
+			default:
+				res.outcome = "exhausted"
+				return res
+			}
+		}
+		if vetoed[picked] {
+			if fs.RecordProfitVeto(picked) == FailoverExhausted {
+				res.outcome = "exhausted"
+				return res
+			}
+			continue
+		}
+		res.outcome = "forwarded"
+		res.forwardedID = picked
+		return res
+	}
+	res.outcome = "budget_exceeded"
+	return res
+}
+
+// TestProfitVetoAfter503DoesNotLivelock 钉死 #4925 引入的活锁回归：
+// 一次真实 503 之后，若调度器持续返回会被利润门否决的账号，
+// 「选号 → 否决 → 排除 → 选号耗尽 → 清空排除 → 睡 2s → 选号」会无限循环，
+// 因为利润否决不推进 SwitchCount，退避条件永远成立。
+func TestProfitVetoAfter503DoesNotLivelock(t *testing.T) {
+	fs := NewFailoverState(10, false)
+	// 已经历一次真实 503（Antigravity 单账号分组 MODEL_CAPACITY_EXHAUSTED 是设计内路径）。
+	fs.LastFailoverErr = newTestFailoverErr(503, false, false)
+	fs.SwitchCount = 1
+	fs.FailedAccountIDs[1] = struct{}{}
+
+	start := time.Now()
+	res := runProfitVetoLoop(t, fs, []int64{1}, map[int64]bool{1: true}, 50)
+	elapsed := time.Since(start)
+
+	require.Equal(t, "exhausted", res.outcome, "整池被利润门否决时必须有限步终止")
+	require.LessOrEqual(t, res.backoffRetries, 1, "被否决的账号不得被退避分支复活并反复重选")
+	require.Less(t, elapsed, 10*time.Second, "不得每 2s 空转一轮")
+}
+
+// TestProfitVetoKeepsBackoffUsefulForHealthyAccount 钉死修复没有削弱既有的
+// 503 退避语义：排除列表里仍有非利润否决的账号时，退避照常清空并重试，
+// 只是被利润门否决的账号不再复活。
+func TestProfitVetoKeepsBackoffUsefulForHealthyAccount(t *testing.T) {
+	fs := NewFailoverState(10, false)
+	fs.LastFailoverErr = newTestFailoverErr(503, false, false)
+	fs.SwitchCount = 1
+	// 账号 1 因真实 503 被排除；账号 2 会被利润门否决。
+	fs.FailedAccountIDs[1] = struct{}{}
+
+	res := runProfitVetoLoop(t, fs, []int64{2, 1}, map[int64]bool{2: true}, 50)
+
+	require.Equal(t, "forwarded", res.outcome)
+	require.Equal(t, int64(1), res.forwardedID, "退避清空后应重新可选账号 1")
+	require.Equal(t, 1, res.backoffRetries, "应发生且只发生一次退避重试")
+	require.Contains(t, fs.FailedAccountIDs, int64(2), "利润否决的账号在退避清空后必须被放回排除集")
+}
+
+// TestProfitVetoAttemptsCapped 钉死没有 503 参与时，大分组整池越线也会在
+// 常数步内终止，而不是把整池逐个选一遍。
+func TestProfitVetoAttemptsCapped(t *testing.T) {
+	fs := NewFailoverState(10, false)
+	pool := make([]int64, 0, 64)
+	vetoed := make(map[int64]bool, 64)
+	for id := int64(1); id <= 64; id++ {
+		pool = append(pool, id)
+		vetoed[id] = true
+	}
+
+	res := runProfitVetoLoop(t, fs, pool, vetoed, 200)
+
+	require.Equal(t, "exhausted", res.outcome)
+	require.Equal(t, maxProfitVetoAttempts, fs.ProfitVetoCount())
+	require.Equal(t, maxProfitVetoAttempts, res.iterations, "达到上限即终止，不应继续遍历候选池")
+}
+
+// TestRecordProfitVetoExcludesAccount 钉死 RecordProfitVeto 仍然把账号加入
+// 调度排除列表（选号入参用的就是 FailedAccountIDs）。
+func TestRecordProfitVetoExcludesAccount(t *testing.T) {
+	fs := NewFailoverState(10, false)
+	require.Equal(t, FailoverContinue, fs.RecordProfitVeto(42))
+	require.Contains(t, fs.FailedAccountIDs, int64(42))
+	require.Equal(t, 1, fs.ProfitVetoCount())
+}
+
+// TestHandleSelectionExhaustedUnaffectedWithoutProfitVeto 钉死未启用利润控制的
+// 请求（无任何利润否决）走的仍是原有退避语义：清空排除列表并重试。
+func TestHandleSelectionExhaustedUnaffectedWithoutProfitVeto(t *testing.T) {
+	fs := NewFailoverState(3, false)
+	fs.LastFailoverErr = newTestFailoverErr(503, false, false)
+	fs.SwitchCount = 1
+	fs.FailedAccountIDs[100] = struct{}{}
+
+	require.Equal(t, FailoverContinue, fs.HandleSelectionExhausted(context.Background()))
+	require.Empty(t, fs.FailedAccountIDs, "无利润否决时排除列表应被完全清空")
+}

backend/internal/handler/gateway_handler.go

@@ -196,6 +196,8 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 
 	setOpsRequestContext(c, reqModel, reqStream)
 	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(reqStream, false)))
+	pricingCtx, pricingAt := service.WithGatewayTokenRequestPricing(c.Request.Context())
+	c.Request = c.Request.WithContext(pricingCtx)
 
 	// 验证 model 必填
 	if reqModel == "" {
@@ -421,8 +423,30 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 				}
 				// Slot acquired: no longer waiting in queue.
 				releaseWait()
-				if err := h.gatewayService.BindStickySession(c.Request.Context(), apiKey.GroupID, sessionKey, account.ID); err != nil {
-					reqLog.Warn("gateway.bind_sticky_session_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+			}
+			// 终检与准入后绑定使用选号结果携带的门（见 responses 同名注释）。
+			admissionCtx := service.ContextWithSelectionProfitGate(c.Request.Context(), selection)
+			latest, vetoed, reason := h.gatewayService.GatewayProfitControlVetoLatest(admissionCtx, account)
+			if vetoed {
+				if accountReleaseFunc != nil {
+					accountReleaseFunc()
+				}
+				reqLog.Debug("gateway.account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+				if fs.RecordProfitVeto(account.ID) == FailoverExhausted {
+					reqLog.Warn("gateway.profit_veto_attempts_exhausted", zap.Int("profit_veto_count", fs.ProfitVetoCount()))
+					markOpsRoutingCapacityLimited(c)
+					h.handleStreamingAwareError(c, http.StatusServiceUnavailable, "api_error", profitVetoExhaustedMessage, streamStarted)
+					return
+				}
+				continue
+			}
+			account = latest
+			selection.Account = latest
+			// 等待路径保持既有 eager 绑定（无门时 helper 直接绑定）；调度器已
+			// 抢槽的直达路径无门时由选号内部绑定，这里只在门下补准入后绑定。
+			if selection.ProfitGateActive() || !selection.Acquired {
+				if err := h.gatewayService.BindStickySessionAfterProfitAdmission(admissionCtx, apiKey.GroupID, sessionKey, account.ID); err != nil {
+					reqLog.Warn("gateway.bind_sticky_session_after_profit_admission_failed", zap.Int64("account_id", account.ID), zap.Error(err))
 				}
 			}
 			// 账号槽位/等待计数需要在超时或断开时安全回收
@@ -535,6 +559,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 			// ForceCacheBilling 提前拍成标量，避免 worker 闭包保活 failover 状态里的响应体。
 			forceCacheBilling := fs.ForceCacheBilling
 			quotaPlatform := service.QuotaPlatform(c.Request.Context(), apiKey)
+			sessionID := service.ExtractClientSessionID(c)
 			h.submitUsageRecordTask(c.Request.Context(), func(ctx context.Context) {
 				if err := h.gatewayService.RecordUsage(ctx, &service.RecordUsageInput{
 					Result:             result,
@@ -543,10 +568,12 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 					User:               apiKey.User,
 					Account:            account,
 					Subscription:       subscription,
+					PricingAt:          pricingAt,
 					InboundEndpoint:    inboundEndpoint,
 					UpstreamEndpoint:   upstreamEndpoint,
 					UserAgent:          userAgent,
 					IPAddress:          clientIP,
+					SessionID:          sessionID,
 					RequestPayloadHash: requestPayloadHash,
 					ForceCacheBilling:  forceCacheBilling,
 					APIKeyService:      h.apiKeyService,
@@ -719,12 +746,30 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 				}
 				// Slot acquired: no longer waiting in queue.
 				releaseWait()
-				reqLog.Info("sticky.bind_after_wait",
-					zap.String("session_key", sessionKey),
-					zap.Int64("account_id", account.ID),
-				)
-				if err := h.gatewayService.BindStickySession(c.Request.Context(), currentAPIKey.GroupID, sessionKey, account.ID); err != nil {
-					reqLog.Warn("gateway.bind_sticky_session_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+			}
+			// 终检与准入后绑定使用选号结果携带的门（见 responses 同名注释）。
+			admissionCtx := service.ContextWithSelectionProfitGate(c.Request.Context(), selection)
+			latest, vetoed, reason := h.gatewayService.GatewayProfitControlVetoLatest(admissionCtx, account)
+			if vetoed {
+				if accountReleaseFunc != nil {
+					accountReleaseFunc()
+				}
+				reqLog.Debug("gateway.account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+				if fs.RecordProfitVeto(account.ID) == FailoverExhausted {
+					reqLog.Warn("gateway.profit_veto_attempts_exhausted", zap.Int("profit_veto_count", fs.ProfitVetoCount()))
+					markOpsRoutingCapacityLimited(c)
+					h.handleStreamingAwareError(c, http.StatusServiceUnavailable, "api_error", profitVetoExhaustedMessage, streamStarted)
+					return
+				}
+				continue
+			}
+			account = latest
+			selection.Account = latest
+			// 等待路径保持既有 eager 绑定（无门时 helper 直接绑定）；调度器已
+			// 抢槽的直达路径无门时由选号内部绑定，这里只在门下补准入后绑定。
+			if selection.ProfitGateActive() || !selection.Acquired {
+				if err := h.gatewayService.BindStickySessionAfterProfitAdmission(admissionCtx, currentAPIKey.GroupID, sessionKey, account.ID); err != nil {
+					reqLog.Warn("gateway.bind_sticky_session_after_profit_admission_failed", zap.Int64("account_id", account.ID), zap.Error(err))
 				}
 			}
 			// 账号槽位/等待计数需要在超时或断开时安全回收
@@ -825,6 +870,66 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 			if accountReleaseFunc != nil {
 				accountReleaseFunc()
 			}
+
+			// 提交 usage 记录。成功路径与"流中断但 Forward 已观测到 usage 的部分结果"
+			// 错误路径共用：后者若不入账，上游已计量的请求会完全漏记漏计费（#5148）。
+			submitForwardUsage := func(result *service.ForwardResult) {
+				// 捕获请求信息（用于异步记录，避免在 goroutine 中访问 gin.Context）
+				userAgent := c.GetHeader("User-Agent")
+				clientIP := ip.GetClientIP(c)
+				// Forward 内部可能继续改写 body，usage 去重指纹必须使用最终上游接受的当前 body。
+				requestPayloadHash := service.HashUsageRequestPayload(attemptParsedReq.Body.Bytes())
+				inboundEndpoint := GetInboundEndpoint(c)
+				upstreamEndpoint := GetUpstreamEndpoint(c, account.Platform)
+
+				if result.ReasoningEffort == nil {
+					result.ReasoningEffort = service.NormalizeClaudeOutputEffort(attemptParsedReq.OutputEffort)
+				}
+				// 同上（重试路径中的对称填充）。详见非重试路径同名注释。
+				if result.ReasoningEffort == nil && attemptParsedReq.ThinkingEnabled {
+					protocolModel := result.UpstreamModel
+					if protocolModel == "" {
+						protocolModel = result.Model
+					}
+					result.ReasoningEffort = service.DefaultEffortForThinkingEnabled(protocolModel)
+				}
+
+				// 使用量记录通过有界 worker 池提交，避免请求热路径创建无界 goroutine。
+				// ForceCacheBilling 提前拍成标量，避免 worker 闭包保活 failover 状态里的响应体。
+				forceCacheBilling := fs.ForceCacheBilling
+				quotaPlatform := service.QuotaPlatform(c.Request.Context(), currentAPIKey)
+				sessionID := service.ExtractClientSessionID(c)
+				h.submitUsageRecordTask(c.Request.Context(), func(ctx context.Context) {
+					if err := h.gatewayService.RecordUsage(ctx, &service.RecordUsageInput{
+						Result:             result,
+						QuotaPlatform:      quotaPlatform,
+						APIKey:             currentAPIKey,
+						User:               currentAPIKey.User,
+						Account:            account,
+						Subscription:       currentSubscription,
+						PricingAt:          pricingAt,
+						InboundEndpoint:    inboundEndpoint,
+						UpstreamEndpoint:   upstreamEndpoint,
+						UserAgent:          userAgent,
+						IPAddress:          clientIP,
+						SessionID:          sessionID,
+						RequestPayloadHash: requestPayloadHash,
+						ForceCacheBilling:  forceCacheBilling,
+						APIKeyService:      h.apiKeyService,
+						ChannelUsageFields: clientRequestedUsageFields(c, channelMapping, reqModel, result.UpstreamModel),
+					}); err != nil {
+						logger.L().With(
+							zap.String("component", "handler.gateway.messages"),
+							zap.Int64("user_id", subject.UserID),
+							zap.Int64("api_key_id", currentAPIKey.ID),
+							zap.Any("group_id", currentAPIKey.GroupID),
+							zap.String("model", reqModel),
+							zap.Int64("account_id", account.ID),
+						).Error("gateway.record_usage_failed", zap.Error(err))
+					}
+				})
+			}
+
 			if err != nil {
 				// Beta policy block: return 400 immediately, no failover
 				var betaBlockedErr *service.BetaBlockedError
@@ -923,6 +1028,12 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 					forwardFailedFields = append(forwardFailedFields, zap.Int64p("proxy_id", account.ProxyID))
 				}
 				reqLog.Error("gateway.forward_failed", forwardFailedFields...)
+				// Forward 与错误一起返回的部分结果：流中断前上游已计量的 usage 照常入账，
+				// 避免上游已产生消耗的请求完全漏记（#5148）。failover 错误恒定 result=nil，
+				// 不会走到这里重复计费。
+				if result != nil {
+					submitForwardUsage(result)
+				}
 				return
 			}
 
@@ -946,57 +1057,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 				}
 			}
 
-			// 捕获请求信息（用于异步记录，避免在 goroutine 中访问 gin.Context）
-			userAgent := c.GetHeader("User-Agent")
-			clientIP := ip.GetClientIP(c)
-			// Forward 内部可能继续改写 body，usage 去重指纹必须使用最终上游接受的当前 body。
-			requestPayloadHash := service.HashUsageRequestPayload(attemptParsedReq.Body.Bytes())
-			inboundEndpoint := GetInboundEndpoint(c)
-			upstreamEndpoint := GetUpstreamEndpoint(c, account.Platform)
-
-			if result.ReasoningEffort == nil {
-				result.ReasoningEffort = service.NormalizeClaudeOutputEffort(attemptParsedReq.OutputEffort)
-			}
-			// 同上（重试路径中的对称填充）。详见非重试路径同名注释。
-			if result.ReasoningEffort == nil && attemptParsedReq.ThinkingEnabled {
-				protocolModel := result.UpstreamModel
-				if protocolModel == "" {
-					protocolModel = result.Model
-				}
-				result.ReasoningEffort = service.DefaultEffortForThinkingEnabled(protocolModel)
-			}
-
-			// 使用量记录通过有界 worker 池提交，避免请求热路径创建无界 goroutine。
-			// ForceCacheBilling 提前拍成标量，避免 worker 闭包保活 failover 状态里的响应体。
-			forceCacheBilling := fs.ForceCacheBilling
-			quotaPlatform := service.QuotaPlatform(c.Request.Context(), currentAPIKey)
-			h.submitUsageRecordTask(c.Request.Context(), func(ctx context.Context) {
-				if err := h.gatewayService.RecordUsage(ctx, &service.RecordUsageInput{
-					Result:             result,
-					QuotaPlatform:      quotaPlatform,
-					APIKey:             currentAPIKey,
-					User:               currentAPIKey.User,
-					Account:            account,
-					Subscription:       currentSubscription,
-					InboundEndpoint:    inboundEndpoint,
-					UpstreamEndpoint:   upstreamEndpoint,
-					UserAgent:          userAgent,
-					IPAddress:          clientIP,
-					RequestPayloadHash: requestPayloadHash,
-					ForceCacheBilling:  forceCacheBilling,
-					APIKeyService:      h.apiKeyService,
-					ChannelUsageFields: clientRequestedUsageFields(c, channelMapping, reqModel, result.UpstreamModel),
-				}); err != nil {
-					logger.L().With(
-						zap.String("component", "handler.gateway.messages"),
-						zap.Int64("user_id", subject.UserID),
-						zap.Int64("api_key_id", currentAPIKey.ID),
-						zap.Any("group_id", currentAPIKey.GroupID),
-						zap.String("model", reqModel),
-						zap.Int64("account_id", account.ID),
-					).Error("gateway.record_usage_failed", zap.Error(err))
-				}
-			})
+			submitForwardUsage(result)
 			return
 		}
 		if !retryWithFallback {
@@ -2122,22 +2183,22 @@ func sendMockInterceptStream(c *gin.Context, model string, interceptType Interce
 
 	switch interceptType {
 	case InterceptTypeSuggestionMode:
-		msgID = "msg_mock_suggestion"
+		msgID = generateRealisticMsgID()
 		outputTokens = 1
 		textDeltas = []string{""} // 空内容
 	default: // InterceptTypeWarmup
-		msgID = "msg_mock_warmup"
+		msgID = generateRealisticMsgID()
 		outputTokens = 2
 		textDeltas = []string{"New", " Conversation"}
 	}
 
-	// Build message_start event with fixed schema.
-	messageStartJSON := `{"type":"message_start","message":{"id":` + strconv.Quote(msgID) + `,"type":"message","role":"assistant","model":` + strconv.Quote(model) + `,"content":[],"stop_reason":null,"stop_sequence":null,"usage":{"input_tokens":10,"output_tokens":0}}}`
+	// Build message_start event — field order matches real Anthropic API response.
+	messageStartJSON := `{"type":"message_start","message":{"model":` + strconv.Quote(model) + `,"id":` + strconv.Quote(msgID) + `,"type":"message","role":"assistant","content":[],"stop_reason":null,"stop_sequence":null,"stop_details":null,"usage":{"input_tokens":10,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0}}}`
 
 	// Build events
 	events := []string{
 		`event: message_start` + "\n" + `data: ` + string(messageStartJSON),
-		`event: content_block_start` + "\n" + `data: {"content_block":{"text":"","type":"text"},"index":0,"type":"content_block_start"}`,
+		`event: content_block_start` + "\n" + `data: {"type":"content_block_start","index":0,"content_block":{"type":"text","text":""}}`,
 	}
 
 	// Add text deltas
@@ -2147,7 +2208,7 @@ func sendMockInterceptStream(c *gin.Context, model string, interceptType Interce
 	}
 
 	// Add final events
-	messageDeltaJSON := `{"type":"message_delta","delta":{"stop_reason":"end_turn","stop_sequence":null},"usage":{"input_tokens":10,"output_tokens":` + strconv.Itoa(outputTokens) + `}}`
+	messageDeltaJSON := `{"type":"message_delta","delta":{"stop_reason":"end_turn","stop_sequence":null,"stop_details":null},"usage":{"output_tokens":` + strconv.Itoa(outputTokens) + `}}`
 
 	events = append(events,
 		`event: content_block_stop`+"\n"+`data: {"index":0,"type":"content_block_stop"}`,
@@ -2162,20 +2223,20 @@ func sendMockInterceptStream(c *gin.Context, model string, interceptType Interce
 	}
 }
 
-// generateRealisticMsgID 生成仿真的消息 ID（msg_bdrk_XXXXXXX 格式）
-// 格式与 Claude API 真实响应一致，24 位随机字母数字
+// generateRealisticMsgID 生成仿真的消息 ID（msg_01XXXXXXX 格式）
+// 格式与 Anthropic API 官方响应一致：msg_01 + 22 位 Base62 随机字符
 func generateRealisticMsgID() string {
 	const charset = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"
-	const idLen = 24
+	const idLen = 22
 	randomBytes := make([]byte, idLen)
 	if _, err := rand.Read(randomBytes); err != nil {
-		return fmt.Sprintf("msg_bdrk_%d", time.Now().UnixNano())
+		return fmt.Sprintf("msg_01%d", time.Now().UnixNano())
 	}
 	b := make([]byte, idLen)
 	for i := range b {
 		b[i] = charset[int(randomBytes[i])%len(charset)]
 	}
-	return "msg_bdrk_" + string(b)
+	return "msg_01" + string(b)
 }
 
 // sendMockInterceptResponse 发送非流式 mock 响应（用于请求拦截）
@@ -2185,7 +2246,7 @@ func sendMockInterceptResponse(c *gin.Context, model string, interceptType Inter
 
 	switch interceptType {
 	case InterceptTypeSuggestionMode:
-		msgID = "msg_mock_suggestion"
+		msgID = generateRealisticMsgID()
 		text = ""
 		outputTokens = 1
 		stopReason = "end_turn"
@@ -2195,13 +2256,13 @@ func sendMockInterceptResponse(c *gin.Context, model string, interceptType Inter
 		outputTokens = 1
 		stopReason = "max_tokens" // max_tokens=1 探测请求的 stop_reason 应为 max_tokens
 	default: // InterceptTypeWarmup
-		msgID = "msg_mock_warmup"
+		msgID = generateRealisticMsgID()
 		text = "New Conversation"
 		outputTokens = 2
 		stopReason = "end_turn"
 	}
 
-	// 构建完整的响应格式（与 Claude API 响应格式一致）
+	// 构建完整的响应格式（与 Anthropic API 官方响应格式一致）
 	response := gin.H{
 		"model":         model,
 		"id":            msgID,
@@ -2210,6 +2271,7 @@ func sendMockInterceptResponse(c *gin.Context, model string, interceptType Inter
 		"content":       []gin.H{{"type": "text", "text": text}},
 		"stop_reason":   stopReason,
 		"stop_sequence": nil,
+		"stop_details":  nil,
 		"usage": gin.H{
 			"input_tokens":                10,
 			"cache_creation_input_tokens": 0,
@@ -2219,7 +2281,6 @@ func sendMockInterceptResponse(c *gin.Context, model string, interceptType Inter
 				"ephemeral_1h_input_tokens": 0,
 			},
 			"output_tokens": outputTokens,
-			"total_tokens":  10 + outputTokens,
 		},
 	}
 
@@ -2332,10 +2393,16 @@ func (h *GatewayHandler) submitUsageRecordTask(parent context.Context, task serv
 	}
 	task = wrapUsageRecordTaskContext(parent, task)
 	if h.usageRecordWorkerPool != nil {
-		h.usageRecordWorkerPool.Submit(task)
-		return
+		if mode := h.usageRecordWorkerPool.Submit(task); mode != service.UsageRecordSubmitModeDroppedStopped {
+			return
+		}
+		// 池已停止（进程关停窗口）：计费任务不能静默丢失，降级为内联同步执行。
+		// 显式配置的 drop/sample 溢出丢弃仍按配置语义保留。
+		logger.L().With(
+			zap.String("component", "handler.gateway.messages"),
+		).Warn("gateway.usage_record_task_stopped_sync_fallback")
 	}
-	// 回退路径：worker 池未注入时同步执行，避免退回到无界 goroutine 模式。
+	// 回退路径：worker 池未注入或已停止时同步执行，避免退回到无界 goroutine 模式。
 	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
 	defer cancel()
 	defer func() {

backend/internal/handler/gateway_handler_chat_completions.go

@@ -93,6 +93,8 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 
 	setOpsRequestContext(c, reqModel, reqStream)
 	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(reqStream, false)))
+	pricingCtx, pricingAt := service.WithGatewayTokenRequestPricing(c.Request.Context())
+	c.Request = c.Request.WithContext(pricingCtx)
 
 	// 解析渠道级模型映射
 	channelMapping, _ := h.gatewayService.ResolveChannelMappingAndRestrict(c.Request.Context(), apiKey.GroupID, reqModel)
@@ -157,7 +159,6 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 	if groupPlatform == service.PlatformGemini && selectionSessionHash != "" {
 		selectionSessionHash = "gemini:" + selectionSessionHash
 	}
-
 	// 3. Account selection + failover loop
 	fs := NewFailoverState(h.maxAccountSwitches, false)
 	if groupPlatform == service.PlatformGemini {
@@ -223,6 +224,28 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 				return
 			}
 		}
+		// 终检与准入后绑定使用选号结果携带的门（见 responses 同名注释）。
+		admissionCtx := service.ContextWithSelectionProfitGate(c.Request.Context(), selection)
+		latest, vetoed, reason := h.gatewayService.GatewayProfitControlVetoLatest(admissionCtx, account)
+		if vetoed {
+			if accountReleaseFunc != nil {
+				accountReleaseFunc()
+			}
+			reqLog.Debug("gateway.cc.account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+			if fs.RecordProfitVeto(account.ID) == FailoverExhausted {
+				reqLog.Warn("gateway.cc.profit_veto_attempts_exhausted", zap.Int("profit_veto_count", fs.ProfitVetoCount()))
+				h.chatCompletionsErrorResponse(c, http.StatusServiceUnavailable, "api_error", profitVetoExhaustedMessage)
+				return
+			}
+			continue
+		}
+		account = latest
+		selection.Account = latest
+		if selection.ProfitGateActive() {
+			if err := h.gatewayService.BindStickySessionAfterProfitAdmission(admissionCtx, apiKey.GroupID, selectionSessionHash, account.ID); err != nil {
+				reqLog.Warn("gateway.cc.bind_sticky_session_after_profit_admission_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+			}
+		}
 		accountReleaseFunc = wrapReleaseOnDone(c.Request.Context(), accountReleaseFunc)
 
 		if groupPlatform == service.PlatformGemini && account.Platform != service.PlatformGemini {
@@ -240,6 +263,7 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 			forwardBody = h.gatewayService.ReplaceModelInBody(body, channelMapping.MappedModel)
 		}
 		var result *service.ForwardResult
+		setActualUpstreamEndpoint(c, "")
 		if account.Platform == service.PlatformGemini {
 			if h.geminiCompatService == nil {
 				h.chatCompletionsErrorResponse(c, http.StatusBadGateway, "upstream_error", "Gemini compatibility service is not configured")
@@ -249,6 +273,16 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 				return
 			}
 			result, err = h.geminiCompatService.ForwardAsChatCompletions(c.Request.Context(), c, account, forwardBody)
+		} else if shouldUseAntigravityCompat(account) {
+			if h.antigravityGatewayService == nil {
+				h.chatCompletionsErrorResponse(c, http.StatusBadGateway, "upstream_error", "Antigravity compatibility service is not configured")
+				if accountReleaseFunc != nil {
+					accountReleaseFunc()
+				}
+				return
+			}
+			setActualUpstreamEndpoint(c, EndpointAntigravityGenerateContent)
+			result, err = h.antigravityGatewayService.ForwardAsChatCompletions(c.Request.Context(), c, account, forwardBody, parsedReq)
 		} else {
 			result, err = h.gatewayService.ForwardAsChatCompletions(c.Request.Context(), c, account, forwardBody, parsedReq)
 		}
@@ -298,6 +332,7 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 		upstreamEndpoint := GetUpstreamEndpoint(c, account.Platform)
 
 		quotaPlatform := service.QuotaPlatform(c.Request.Context(), apiKey)
+		sessionID := service.ExtractClientSessionID(c)
 		h.submitUsageRecordTask(c.Request.Context(), func(ctx context.Context) {
 			if err := h.gatewayService.RecordUsage(ctx, &service.RecordUsageInput{
 				Result:             result,
@@ -306,12 +341,14 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 				User:               apiKey.User,
 				Account:            account,
 				Subscription:       subscription,
+				PricingAt:          pricingAt,
 				InboundEndpoint:    inboundEndpoint,
 				UpstreamEndpoint:   upstreamEndpoint,
 				UserAgent:          userAgent,
 				IPAddress:          clientIP,
 				RequestPayloadHash: requestPayloadHash,
 				APIKeyService:      h.apiKeyService,
+				SessionID:          sessionID,
 				ChannelUsageFields: clientRequestedUsageFields(c, channelMapping, reqModel, result.UpstreamModel),
 			}); err != nil {
 				reqLog.Error("gateway.cc.record_usage_failed",

backend/internal/handler/gateway_handler_error_fallback_test.go

@@ -8,6 +8,7 @@ import (
 	"strings"
 	"testing"
 
+	"github.com/Wei-Shaw/sub2api/internal/service"
 	"github.com/gin-gonic/gin"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
@@ -53,6 +54,22 @@ func TestGatewayEnsureForwardErrorResponse_AppendsSSEAfterWritten(t *testing.T)
 	assert.Contains(t, w.Body.String(), `data: {"type":"error"`)
 }
 
+func TestGatewayEnsureForwardErrorResponse_SkipsCommittedSSEError(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	w := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(w)
+	c.Request = httptest.NewRequest(http.MethodPost, EndpointResponses, nil)
+	c.Header("Content-Type", "text/event-stream")
+	_, _ = c.Writer.WriteString("event: error\ndata: {\"type\":\"error\"}\n\n")
+	service.MarkResponseCommitted(c)
+
+	h := &GatewayHandler{}
+	wrote := h.ensureForwardErrorResponse(c, true)
+
+	require.False(t, wrote)
+	require.Equal(t, 1, strings.Count(w.Body.String(), "event: error"))
+}
+
 // case B 回归：Anthropic-backed /responses，Writer 已被写过时
 // ensureForwardErrorResponse 仍要发 response.failed。
 func TestGatewayEnsureForwardErrorResponse_ResponsesRouteAfterWrittenEmitsResponseFailed(t *testing.T) {

backend/internal/handler/gateway_handler_intercept_test.go

@@ -49,7 +49,7 @@ func TestSendMockInterceptResponse_MaxTokensOneHaiku(t *testing.T) {
 
 	id, ok := response["id"].(string)
 	require.True(t, ok)
-	require.True(t, strings.HasPrefix(id, "msg_bdrk_"))
+	require.True(t, strings.HasPrefix(id, "msg_01"))
 
 	content, ok := response["content"].([]any)
 	require.True(t, ok)

backend/internal/handler/gateway_handler_responses.go

@@ -90,9 +90,15 @@ func (h *GatewayHandler) Responses(c *gin.Context) {
 	setOpsRequestContext(c, reqModel, reqStream)
 	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(reqStream, false)))
 	requestCtx := c.Request.Context()
+	// 定价上下文无条件装配：/v1/responses 是 token 计费端点，声明生图工具的
+	// 混合请求同样按 token 计费（外加图片部分），其 token 利润保护不因请求体
+	// 里的任何工具声明（含 Codex 被动 image_gen namespace）而关闭。生图意图
+	// 仅用于能力路由与图片计费；独立图片/视频端点才在利润门范围之外。
+	requestCtx, pricingAt := service.WithGatewayTokenRequestPricing(requestCtx)
 	if service.IsImageGenerationIntentForPlatform("/v1/responses", reqModel, body, openAICompatibleRequestPlatform(c.Request.Context(), apiKey)) {
 		requestCtx = service.WithOpenAIImageGenerationIntent(requestCtx)
 	}
+	c.Request = c.Request.WithContext(requestCtx)
 
 	// 解析渠道级模型映射
 	channelMapping, _ := h.gatewayService.ResolveChannelMappingAndRestrict(requestCtx, apiKey.GroupID, reqModel)
@@ -220,6 +226,30 @@ func (h *GatewayHandler) Responses(c *gin.Context) {
 				return
 			}
 		}
+		// 终检与准入后绑定必须使用选号结果携带的门：门安装在调度栈的局部
+		// ctx 上（composite/fallback 还可能解析出与入口分组不同的门），直接用
+		// requestCtx 会退化为空操作。
+		admissionCtx := service.ContextWithSelectionProfitGate(requestCtx, selection)
+		latest, vetoed, reason := h.gatewayService.GatewayProfitControlVetoLatest(admissionCtx, account)
+		if vetoed {
+			if accountReleaseFunc != nil {
+				accountReleaseFunc()
+			}
+			reqLog.Debug("gateway.responses.account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+			if fs.RecordProfitVeto(account.ID) == FailoverExhausted {
+				reqLog.Warn("gateway.responses.profit_veto_attempts_exhausted", zap.Int("profit_veto_count", fs.ProfitVetoCount()))
+				h.responsesErrorResponse(c, http.StatusServiceUnavailable, "api_error", profitVetoExhaustedMessage)
+				return
+			}
+			continue
+		}
+		account = latest
+		selection.Account = latest
+		if selection.ProfitGateActive() {
+			if err := h.gatewayService.BindStickySessionAfterProfitAdmission(admissionCtx, apiKey.GroupID, sessionHash, account.ID); err != nil {
+				reqLog.Warn("gateway.responses.bind_sticky_session_after_profit_admission_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+			}
+		}
 		accountReleaseFunc = wrapReleaseOnDone(c.Request.Context(), accountReleaseFunc)
 
 		// 5. Forward request
@@ -228,7 +258,21 @@ func (h *GatewayHandler) Responses(c *gin.Context) {
 		if channelMapping.Mapped {
 			forwardBody = h.gatewayService.ReplaceModelInBody(body, channelMapping.MappedModel)
 		}
-		result, err := h.gatewayService.ForwardAsResponses(requestCtx, c, account, forwardBody, parsedReq)
+		var result *service.ForwardResult
+		setActualUpstreamEndpoint(c, "")
+		if shouldUseAntigravityCompat(account) {
+			if h.antigravityGatewayService == nil {
+				h.responsesErrorResponse(c, http.StatusBadGateway, "upstream_error", "Antigravity compatibility service is not configured")
+				if accountReleaseFunc != nil {
+					accountReleaseFunc()
+				}
+				return
+			}
+			setActualUpstreamEndpoint(c, EndpointAntigravityGenerateContent)
+			result, err = h.antigravityGatewayService.ForwardAsResponses(requestCtx, c, account, forwardBody, parsedReq)
+		} else {
+			result, err = h.gatewayService.ForwardAsResponses(requestCtx, c, account, forwardBody, parsedReq)
+		}
 
 		if accountReleaseFunc != nil {
 			accountReleaseFunc()
@@ -276,6 +320,7 @@ func (h *GatewayHandler) Responses(c *gin.Context) {
 		upstreamEndpoint := GetUpstreamEndpoint(c, account.Platform)
 
 		quotaPlatform := service.QuotaPlatform(c.Request.Context(), apiKey)
+		sessionID := service.ExtractClientSessionID(c)
 		h.submitUsageRecordTask(c.Request.Context(), func(ctx context.Context) {
 			if err := h.gatewayService.RecordUsage(ctx, &service.RecordUsageInput{
 				Result:             result,
@@ -284,12 +329,14 @@ func (h *GatewayHandler) Responses(c *gin.Context) {
 				User:               apiKey.User,
 				Account:            account,
 				Subscription:       subscription,
+				PricingAt:          pricingAt,
 				InboundEndpoint:    inboundEndpoint,
 				UpstreamEndpoint:   upstreamEndpoint,
 				UserAgent:          userAgent,
 				IPAddress:          clientIP,
 				RequestPayloadHash: requestPayloadHash,
 				APIKeyService:      h.apiKeyService,
+				SessionID:          sessionID,
 				ChannelUsageFields: clientRequestedUsageFields(c, channelMapping, reqModel, result.UpstreamModel),
 			}); err != nil {
 				reqLog.Error("gateway.responses.record_usage_failed",

backend/internal/handler/gateway_handler_warmup_intercept_unit_test.go

@@ -7,6 +7,7 @@ import (
 	"context"
 	"encoding/json"
 	"net/http/httptest"
+	"strings"
 	"testing"
 	"time"
 
@@ -289,7 +290,7 @@ func TestGatewayHandlerMessages_InterceptWarmup_AntigravityAccount_MixedScheduli
 
 	var resp map[string]any
 	require.NoError(t, json.Unmarshal(rec.Body.Bytes(), &resp))
-	require.Equal(t, "msg_mock_warmup", resp["id"])
+	require.True(t, strings.HasPrefix(resp["id"].(string), "msg_01"))
 	require.Equal(t, "claude-sonnet-4-5", resp["model"])
 
 	content, ok := resp["content"].([]any)
@@ -378,6 +379,6 @@ func TestGatewayHandlerMessages_InterceptWarmup_AntigravityAccount_ForcePlatform
 
 	var resp map[string]any
 	require.NoError(t, json.Unmarshal(rec.Body.Bytes(), &resp))
-	require.Equal(t, "msg_mock_warmup", resp["id"])
+	require.True(t, strings.HasPrefix(resp["id"].(string), "msg_01"))
 	require.Equal(t, "claude-sonnet-4-5", resp["model"])
 }

backend/internal/handler/gemini_v1beta_handler.go

@@ -98,6 +98,12 @@ func (h *GatewayHandler) GeminiV1BetaGetModel(c *gin.Context) {
 		googleError(c, http.StatusBadRequest, "Missing model in URL")
 		return
 	}
+	// 模型名会被拼进上游 URL 的 path，先在入口校验片段合规性，
+	// 见 service/upstream_path_guard.go。
+	if !service.IsSafeGeminiModelPathSegment(modelName) {
+		googleError(c, http.StatusBadRequest, "Invalid model in URL")
+		return
+	}
 	if resolvedModel, ok := service.ResolvedUpstreamModelFromContext(c.Request.Context()); ok && strings.TrimSpace(resolvedModel) != "" {
 		modelName = strings.TrimSpace(resolvedModel)
 	}
@@ -169,6 +175,12 @@ func (h *GatewayHandler) GeminiV1BetaModels(c *gin.Context) {
 		googleError(c, http.StatusNotFound, err.Error())
 		return
 	}
+	// URL 里的模型名最终会被拼进上游 /v1beta/models/{model}:{action}，
+	// 先在入口校验片段合规性，见 service/upstream_path_guard.go。
+	if !service.IsSafeGeminiModelPathSegment(modelName) {
+		googleError(c, http.StatusBadRequest, "Invalid model in URL")
+		return
+	}
 	if resolvedModel, ok := service.ResolvedUpstreamModelFromContext(c.Request.Context()); ok && strings.TrimSpace(resolvedModel) != "" {
 		modelName = strings.TrimSpace(resolvedModel)
 	}
@@ -192,6 +204,8 @@ func (h *GatewayHandler) GeminiV1BetaModels(c *gin.Context) {
 
 	setOpsRequestContext(c, modelName, stream)
 	setOpsEndpointContext(c, "", int16(service.RequestTypeFromLegacy(stream, false)))
+	pricingCtx, pricingAt := service.WithGatewayTokenRequestPricing(c.Request.Context())
+	c.Request = c.Request.WithContext(pricingCtx)
 
 	if decision := h.checkSecurityAudit(c, reqLog, apiKey, authSubject, service.ContentModerationProtocolGemini, modelName, body); decision != nil && !decision.AllowNextStage {
 		googleSecurityAuditError(c, decision)
@@ -454,8 +468,30 @@ func (h *GatewayHandler) GeminiV1BetaModels(c *gin.Context) {
 				geminiConcurrency.DecrementAccountWaitCount(c.Request.Context(), account.ID)
 				accountWaitCounted = false
 			}
-			if err := h.gatewayService.BindStickySession(c.Request.Context(), apiKey.GroupID, sessionKey, account.ID); err != nil {
-				reqLog.Warn("gemini.bind_sticky_session_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+		}
+		// 终检与准入后绑定使用选号结果携带的门（见 responses 同名注释）。
+		admissionCtx := service.ContextWithSelectionProfitGate(c.Request.Context(), selection)
+		latest, vetoed, reason := h.gatewayService.GatewayProfitControlVetoLatest(admissionCtx, account)
+		if vetoed {
+			if accountReleaseFunc != nil {
+				accountReleaseFunc()
+			}
+			reqLog.Debug("gemini.account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+			if fs.RecordProfitVeto(account.ID) == FailoverExhausted {
+				reqLog.Warn("gemini.profit_veto_attempts_exhausted", zap.Int("profit_veto_count", fs.ProfitVetoCount()))
+				markOpsRoutingCapacityLimited(c)
+				googleError(c, http.StatusServiceUnavailable, profitVetoExhaustedMessage)
+				return
+			}
+			continue
+		}
+		account = latest
+		selection.Account = latest
+		// 等待路径保持既有 eager 绑定（无门时 helper 直接绑定）；调度器已抢槽
+		// 的直达路径无门时由选号内部绑定，这里只在门下补准入后绑定。
+		if selection.ProfitGateActive() || !selection.Acquired {
+			if err := h.gatewayService.BindStickySessionAfterProfitAdmission(admissionCtx, apiKey.GroupID, sessionKey, account.ID); err != nil {
+				reqLog.Warn("gemini.bind_sticky_session_after_profit_admission_failed", zap.Int64("account_id", account.ID), zap.Error(err))
 			}
 		}
 		// 账号槽位/等待计数需要在超时或断开时安全回收
@@ -532,6 +568,7 @@ func (h *GatewayHandler) GeminiV1BetaModels(c *gin.Context) {
 		// ForceCacheBilling 提前拍成标量，避免 worker 闭包保活 failover 状态里的响应体。
 		forceCacheBilling := fs.ForceCacheBilling
 		quotaPlatform := service.QuotaPlatform(c.Request.Context(), apiKey)
+		sessionID := service.ExtractClientSessionID(c)
 		h.submitUsageRecordTask(c.Request.Context(), func(ctx context.Context) {
 			if err := h.gatewayService.RecordUsageWithLongContext(ctx, &service.RecordUsageLongContextInput{
 				Result:                result,
@@ -540,6 +577,7 @@ func (h *GatewayHandler) GeminiV1BetaModels(c *gin.Context) {
 				User:                  apiKey.User,
 				Account:               account,
 				Subscription:          subscription,
+				PricingAt:             pricingAt,
 				InboundEndpoint:       inboundEndpoint,
 				UpstreamEndpoint:      upstreamEndpoint,
 				UserAgent:             userAgent,
@@ -549,6 +587,7 @@ func (h *GatewayHandler) GeminiV1BetaModels(c *gin.Context) {
 				LongContextMultiplier: 2.0,    // 超出部分双倍计费
 				ForceCacheBilling:     forceCacheBilling,
 				APIKeyService:         h.apiKeyService,
+				SessionID:             sessionID,
 				ChannelUsageFields:    clientRequestedUsageFields(c, channelMapping, reqModel, result.UpstreamModel),
 			}); err != nil {
 				logger.L().With(

backend/internal/handler/grok_media.go

@@ -177,7 +177,11 @@ func (h *OpenAIGatewayHandler) handleGrokMedia(c *gin.Context, endpoint service.
 			return
 		}
 	}
-	requestCtx := c.Request.Context()
+	// Grok 媒体（图片/视频生成与视频查询）按媒体倍率计费，不在 token 利润门
+	// 范围内：显式豁免，防止 service 层防御性装门按文本 D 误过滤媒体请求，
+	// 也防止已计费的在途视频任务因绑定账号被门排除而查询返回伪 404。
+	requestCtx := service.WithOpenAIProfitControlSuppressed(c.Request.Context())
+	profitVetoCount := 0
 	failedAccountIDs := make(map[int64]struct{})
 	sameAccountRetryCount := make(map[int64]int)
 	var lastFailoverErr *service.UpstreamFailoverError
@@ -293,8 +297,17 @@ func (h *OpenAIGatewayHandler) handleGrokMedia(c *gin.Context, endpoint service.
 		sessionHash = ensureOpenAIPoolModeSessionHash(sessionHash, account)
 		setOpsSelectedAccount(c, account.ID, account.Platform)
 
-		accountReleaseFunc, accountAcquired := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, false, &streamStarted, reqLog)
-		if !accountAcquired {
+		accountReleaseFunc, slotResult := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, false, &streamStarted, reqLog)
+		if slotResult == openAISlotAcquireProfitVetoed {
+			// 媒体路径已显式豁免利润门（suppress 标记），此分支仅防御性兜底，
+			// 同样受否决上限约束。
+			if !recordOpenAIProfitVeto(failedAccountIDs, account.ID, &profitVetoCount) {
+				h.handleOpenAIProfitVetoExhausted(c, streamStarted, reqLog, profitVetoCount)
+				return
+			}
+			continue
+		}
+		if slotResult != openAISlotAcquireOK {
 			return
 		}
 
@@ -465,6 +478,7 @@ func recordGrokMediaUsage(
 ) {
 	userAgent := c.GetHeader("User-Agent")
 	clientIP := ip.GetClientIP(c)
+	sessionID := service.ExtractClientSessionID(c)
 	payloadForHash := body
 	if len(payloadForHash) == 0 && strings.TrimSpace(requestID) != "" {
 		payloadForHash = []byte(requestID)
@@ -493,6 +507,7 @@ func recordGrokMediaUsage(
 			RequestPayloadHash: service.HashUsageRequestPayload(payloadForHash),
 			APIKeyService:      h.apiKeyService,
 			QuotaPlatform:      quotaPlatform,
+			SessionID:          sessionID,
 			ChannelUsageFields: channelUsageFields,
 		}); err != nil {
 			logger.L().With(

backend/internal/handler/handler.go

@@ -58,9 +58,11 @@ type Handlers struct {
 	OpenAIGateway    *OpenAIGatewayHandler
 	Setting          *SettingHandler
 	Totp             *TotpHandler
+	Passkey          *PasskeyHandler
 	Payment          *PaymentHandler
 	PaymentWebhook   *PaymentWebhookHandler
 	AvailableChannel *AvailableChannelHandler
+	ModelPlaza       *ModelPlazaHandler
 	AsyncImage       *AsyncImageHandler
 	BatchImage       *BatchImageHandler
 }

backend/internal/handler/model_plaza_handler.go

@@ -0,0 +1,198 @@
+package handler
+
+import (
+	"log/slog"
+
+	"github.com/Wei-Shaw/sub2api/internal/pkg/response"
+	"github.com/Wei-Shaw/sub2api/internal/server/middleware"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+
+	"github.com/gin-gonic/gin"
+)
+
+// ModelPlazaHandler 处理「模型广场」查询。
+//
+// 广场路由挂 OptionalJWT 中间件：匿名可访问（除非 require_auth 开启），带 token 则
+// 识别用户。可见性规则（橱窗语义，与「可用渠道」的可绑定语义不同）：
+//   - 匿名：仅非专属分组（订阅型照常展示）；
+//   - 登录：非专属分组 + user_allowed_groups 授权的专属分组（不检查订阅有效性）。
+type ModelPlazaHandler struct {
+	channelService *service.ChannelService
+	apiKeyService  *service.APIKeyService
+	settingService *service.SettingService
+}
+
+// NewModelPlazaHandler 创建模型广场 handler。
+func NewModelPlazaHandler(
+	channelService *service.ChannelService,
+	apiKeyService *service.APIKeyService,
+	settingService *service.SettingService,
+) *ModelPlazaHandler {
+	return &ModelPlazaHandler{
+		channelService: channelService,
+		apiKeyService:  apiKeyService,
+		settingService: settingService,
+	}
+}
+
+// modelPlazaOfficialPricing LiteLLM 官方参考价（USD per token）。
+type modelPlazaOfficialPricing struct {
+	InputPrice        *float64 `json:"input_price"`
+	OutputPrice       *float64 `json:"output_price"`
+	CacheWritePrice   *float64 `json:"cache_write_price"`
+	CacheWrite1hPrice *float64 `json:"cache_write_1h_price,omitempty"`
+	CacheReadPrice    *float64 `json:"cache_read_price"`
+}
+
+// modelPlazaModel 广场模型条目：渠道定价（白名单形态）+ 官方参考价。
+type modelPlazaModel struct {
+	Name            string                     `json:"name"`
+	Platform        string                     `json:"platform"`
+	Pricing         *userSupportedModelPricing `json:"pricing"`
+	OfficialPricing *modelPlazaOfficialPricing `json:"official_pricing"`
+}
+
+// modelPlazaGroup 广场分组条目（白名单字段）。
+type modelPlazaGroup struct {
+	ID                 int64             `json:"id"`
+	Name               string            `json:"name"`
+	Description        string            `json:"description"`
+	Platform           string            `json:"platform"`
+	SubscriptionType   string            `json:"subscription_type"`
+	RateMultiplier     float64           `json:"rate_multiplier"`
+	UserRateMultiplier *float64          `json:"user_rate_multiplier,omitempty"`
+	PeakRateEnabled    bool              `json:"peak_rate_enabled"`
+	PeakStart          string            `json:"peak_start"`
+	PeakEnd            string            `json:"peak_end"`
+	PeakRateMultiplier float64           `json:"peak_rate_multiplier"`
+	IsExclusive        bool              `json:"is_exclusive"`
+	Models             []modelPlazaModel `json:"models"`
+}
+
+// modelPlazaResponse 广场页响应。
+type modelPlazaResponse struct {
+	Description string            `json:"description"`
+	Groups      []modelPlazaGroup `json:"groups"`
+}
+
+// Get 返回模型广场数据。
+// GET /api/v1/model-plaza
+func (h *ModelPlazaHandler) Get(c *gin.Context) {
+	if h.settingService == nil {
+		response.NotFound(c, "Model plaza is not enabled")
+		return
+	}
+	rt := h.settingService.GetModelPlazaRuntime(c.Request.Context())
+	if !rt.Enabled {
+		response.NotFound(c, "Model plaza is not enabled")
+		return
+	}
+
+	subject, authed := middleware.GetAuthSubjectFromContext(c)
+	if rt.RequireAuth && !authed {
+		response.Unauthorized(c, "Authentication required")
+		return
+	}
+
+	groups, err := h.channelService.ListPlazaGroups(c.Request.Context())
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	// allowedExclusive == nil 表示匿名；登录用户恒为非 nil（可能为空集合）。
+	var allowedExclusive map[int64]struct{}
+	var userRates map[int64]float64
+	if authed {
+		allowedExclusive, err = h.apiKeyService.GetUserAllowedGroupIDSet(c.Request.Context(), subject.UserID)
+		if err != nil {
+			// 可见性数据拿不到时不能静默降级成匿名视图（会错漏专属分组），直接报错。
+			response.ErrorFrom(c, err)
+			return
+		}
+		userRates, err = h.apiKeyService.GetUserGroupRates(c.Request.Context(), subject.UserID)
+		if err != nil {
+			// 专属倍率仅是展示增强，失败降级为分组默认倍率。
+			slog.Warn("model_plaza_user_rates_failed", "error", err, "user_id", subject.UserID)
+			userRates = nil
+		}
+	}
+
+	visible := filterPlazaVisibleGroups(groups, allowedExclusive)
+
+	out := make([]modelPlazaGroup, 0, len(visible))
+	for i := range visible {
+		out = append(out, toModelPlazaGroupDTO(&visible[i], userRates))
+	}
+	response.Success(c, modelPlazaResponse{
+		Description: rt.Description,
+		Groups:      out,
+	})
+}
+
+// filterPlazaVisibleGroups 按登录态裁剪分组可见性。
+// allowedExclusive == nil 表示匿名（仅非专属）；非 nil 表示登录（非专属 + 授权专属）。
+func filterPlazaVisibleGroups(
+	groups []service.PlazaGroup,
+	allowedExclusive map[int64]struct{},
+) []service.PlazaGroup {
+	visible := make([]service.PlazaGroup, 0, len(groups))
+	for _, g := range groups {
+		if g.IsExclusive {
+			if allowedExclusive == nil {
+				continue
+			}
+			if _, ok := allowedExclusive[g.ID]; !ok {
+				continue
+			}
+		}
+		visible = append(visible, g)
+	}
+	return visible
+}
+
+// toModelPlazaGroupDTO 将 service 层广场分组映射为白名单 DTO,并合并用户专属倍率。
+func toModelPlazaGroupDTO(g *service.PlazaGroup, userRates map[int64]float64) modelPlazaGroup {
+	models := make([]modelPlazaModel, 0, len(g.Models))
+	for i := range g.Models {
+		m := &g.Models[i]
+		models = append(models, modelPlazaModel{
+			Name:            m.Name,
+			Platform:        m.Platform,
+			Pricing:         toUserPricing(m.Pricing),
+			OfficialPricing: toModelPlazaOfficialPricing(m.OfficialPricing),
+		})
+	}
+	dto := modelPlazaGroup{
+		ID:                 g.ID,
+		Name:               g.Name,
+		Description:        g.Description,
+		Platform:           g.Platform,
+		SubscriptionType:   g.SubscriptionType,
+		RateMultiplier:     g.RateMultiplier,
+		PeakRateEnabled:    g.PeakRateEnabled,
+		PeakStart:          g.PeakStart,
+		PeakEnd:            g.PeakEnd,
+		PeakRateMultiplier: g.PeakRateMultiplier,
+		IsExclusive:        g.IsExclusive,
+		Models:             models,
+	}
+	if rate, ok := userRates[g.ID]; ok {
+		dto.UserRateMultiplier = &rate
+	}
+	return dto
+}
+
+// toModelPlazaOfficialPricing 转换官方参考价；nil 透传（前端显示 "-"）。
+func toModelPlazaOfficialPricing(p *service.PlazaOfficialPricing) *modelPlazaOfficialPricing {
+	if p == nil {
+		return nil
+	}
+	return &modelPlazaOfficialPricing{
+		InputPrice:        p.InputPrice,
+		OutputPrice:       p.OutputPrice,
+		CacheWritePrice:   p.CacheWritePrice,
+		CacheWrite1hPrice: p.CacheWrite1hPrice,
+		CacheReadPrice:    p.CacheReadPrice,
+	}
+}

backend/internal/handler/model_plaza_handler_test.go

@@ -0,0 +1,126 @@
+//go:build unit
+
+package handler
+
+import (
+	"encoding/json"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/require"
+)
+
+func plazaGroups() []service.PlazaGroup {
+	return []service.PlazaGroup{
+		{ID: 1, Name: "public-standard", Platform: "anthropic", SubscriptionType: "standard", RateMultiplier: 1},
+		{ID: 2, Name: "exclusive-a", Platform: "anthropic", IsExclusive: true, RateMultiplier: 0.5},
+		{ID: 3, Name: "public-subscription", Platform: "openai", SubscriptionType: "subscription", RateMultiplier: 1},
+		{ID: 4, Name: "exclusive-b", Platform: "openai", IsExclusive: true, RateMultiplier: 0.8},
+	}
+}
+
+func TestFilterPlazaVisibleGroups_AnonymousSeesOnlyNonExclusive(t *testing.T) {
+	// 匿名(allowedExclusive == nil):仅非专属分组;订阅型公开分组照常可见(橱窗语义)。
+	visible := filterPlazaVisibleGroups(plazaGroups(), nil)
+	require.Len(t, visible, 2)
+	ids := []int64{visible[0].ID, visible[1].ID}
+	require.ElementsMatch(t, []int64{1, 3}, ids)
+}
+
+func TestFilterPlazaVisibleGroups_AuthedSeesGrantedExclusive(t *testing.T) {
+	// 登录:非专属 + 授权的专属;未授权的专属仍不可见。
+	allowed := map[int64]struct{}{2: {}}
+	visible := filterPlazaVisibleGroups(plazaGroups(), allowed)
+	require.Len(t, visible, 3)
+	ids := make([]int64, 0, len(visible))
+	for _, g := range visible {
+		ids = append(ids, g.ID)
+	}
+	require.ElementsMatch(t, []int64{1, 2, 3}, ids)
+}
+
+func TestFilterPlazaVisibleGroups_AuthedEmptySetSeesNoExclusive(t *testing.T) {
+	// 登录但无任何专属授权(空集合,非 nil):与匿名同样只见非专属,
+	// 但语义区分要保持——空集合不能被当作 nil 匿名分支。
+	visible := filterPlazaVisibleGroups(plazaGroups(), map[int64]struct{}{})
+	require.Len(t, visible, 2)
+}
+
+func TestModelPlazaHandler_NilSettingServiceFailsClosed404(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	h := &ModelPlazaHandler{} // settingService == nil → fail-closed
+	w := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(w)
+	c.Request = httptest.NewRequest(http.MethodGet, "/api/v1/model-plaza", nil)
+
+	h.Get(c)
+
+	require.Equal(t, http.StatusNotFound, w.Code)
+}
+
+func TestToModelPlazaGroupDTO_UserRateAndFieldWhitelist(t *testing.T) {
+	g := service.PlazaGroup{
+		ID: 2, Name: "vip", Description: "d", Platform: "anthropic",
+		SubscriptionType: "standard", RateMultiplier: 1, IsExclusive: true,
+		Models: []service.PlazaModel{{
+			Name:     "claude-sonnet",
+			Platform: "anthropic",
+			Pricing: &service.ChannelModelPricing{
+				BillingMode: service.BillingModeToken,
+				InputPrice:  testPtr(3e-6),
+			},
+			OfficialPricing: &service.PlazaOfficialPricing{
+				InputPrice:     testPtr(3e-6),
+				CacheReadPrice: testPtr(3e-7),
+			},
+		}},
+	}
+
+	// 有专属倍率:user_rate_multiplier 序列化输出
+	dto := toModelPlazaGroupDTO(&g, map[int64]float64{2: 0.5})
+	raw, err := json.Marshal(dto)
+	require.NoError(t, err)
+	var decoded map[string]any
+	require.NoError(t, json.Unmarshal(raw, &decoded))
+
+	for _, key := range []string{
+		"id", "name", "description", "platform", "subscription_type",
+		"rate_multiplier", "user_rate_multiplier", "is_exclusive", "models",
+		"peak_rate_enabled", "peak_start", "peak_end", "peak_rate_multiplier",
+	} {
+		_, exists := decoded[key]
+		require.Truef(t, exists, "plaza group DTO must expose %q", key)
+	}
+	require.InDelta(t, 0.5, decoded["user_rate_multiplier"].(float64), 1e-9)
+
+	// 模型条目:pricing + official_pricing 并存;official 缺失字段输出 null 而非省略
+	models := decoded["models"].([]any)
+	require.Len(t, models, 1)
+	model := models[0].(map[string]any)
+	require.Contains(t, model, "pricing")
+	require.Contains(t, model, "official_pricing")
+	official := model["official_pricing"].(map[string]any)
+	require.Contains(t, official, "input_price")
+	require.Contains(t, official, "cache_read_price")
+	_, has1h := official["cache_write_1h_price"]
+	require.False(t, has1h, "1h 缓存写价为 nil 时应 omitempty")
+
+	// 无专属倍率:user_rate_multiplier 整个字段省略
+	dtoNoRate := toModelPlazaGroupDTO(&g, nil)
+	rawNoRate, err := json.Marshal(dtoNoRate)
+	require.NoError(t, err)
+	var decodedNoRate map[string]any
+	require.NoError(t, json.Unmarshal(rawNoRate, &decodedNoRate))
+	_, hasRate := decodedNoRate["user_rate_multiplier"]
+	require.False(t, hasRate, "无专属倍率时 user_rate_multiplier 应 omitempty")
+}
+
+func TestToModelPlazaOfficialPricing_NilPassthrough(t *testing.T) {
+	require.Nil(t, toModelPlazaOfficialPricing(nil))
+}
+
+func testPtr(v float64) *float64 { return &v }

backend/internal/handler/openai_alpha_search.go

@@ -107,12 +107,18 @@ func (h *OpenAIGatewayHandler) AlphaSearch(c *gin.Context) {
 
 	searchID := strings.TrimSpace(gjson.GetBytes(body, "id").String())
 	sessionHash := h.gatewayService.GenerateSessionHashWithFallback(c, nil, searchID)
+	profitVetoCount := 0
 	failedAccountIDs := make(map[int64]struct{})
 	var lastFailoverErr *service.UpstreamFailoverError
 	switchCount := 0
 	var oauth429FailoverState service.OpenAIOAuth429FailoverState
 	routingStart := time.Now()
 
+	// 分组利润控制：alpha search 文本入口请求级装门并固定 pricingAt
+	//（记录路径经 service.OpenAIPricingAtFromContext 从请求 ctx 回读）。
+	asPricingCtx, _ := h.gatewayService.WithOpenAIRequestPricingContext(c.Request.Context(), apiKey.GroupID)
+	c.Request = c.Request.WithContext(asPricingCtx)
+
 	for {
 		selection, _, err := h.gatewayService.SelectAccountWithSchedulerForCapability(
 			c.Request.Context(),
@@ -151,8 +157,16 @@ func (h *OpenAIGatewayHandler) AlphaSearch(c *gin.Context) {
 
 		account := selection.Account
 		setOpsSelectedAccount(c, account.ID, account.Platform)
-		accountRelease, acquired := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, false, &streamStarted, reqLog)
-		if !acquired {
+		accountRelease, slotResult := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, false, &streamStarted, reqLog)
+		if slotResult == openAISlotAcquireProfitVetoed {
+			// 利润终检否决：排除该账号重新选号；否决次数达上限则按无可用账号终止。
+			if !recordOpenAIProfitVeto(failedAccountIDs, account.ID, &profitVetoCount) {
+				h.handleOpenAIProfitVetoExhausted(c, streamStarted, reqLog, profitVetoCount)
+				return
+			}
+			continue
+		}
+		if slotResult != openAISlotAcquireOK {
 			return
 		}
 		service.SetOpsLatencyMs(c, service.OpsRoutingLatencyMsKey, time.Since(routingStart).Milliseconds())
@@ -233,6 +247,7 @@ func (h *OpenAIGatewayHandler) recordAlphaSearchUsage(
 ) {
 	userAgent := c.GetHeader("User-Agent")
 	clientIP := ip.GetClientIP(c)
+	sessionID := service.ExtractClientSessionID(c)
 	requestPayloadHash := service.HashUsageRequestPayload(body)
 	inboundEndpoint := GetInboundEndpoint(c)
 	upstreamEndpoint := GetUpstreamEndpoint(c, account.Platform)
@@ -252,7 +267,9 @@ func (h *OpenAIGatewayHandler) recordAlphaSearchUsage(
 			RequestPayloadHash: requestPayloadHash,
 			APIKeyService:      h.apiKeyService,
 			QuotaPlatform:      quotaPlatform,
+			SessionID:          sessionID,
 			ChannelUsageFields: channelMapping.ToUsageFields(requestedModel, result.UpstreamModel),
+			PricingAt:          service.OpenAIPricingAtFromContext(c.Request.Context()),
 		}); err != nil {
 			logger.L().With(
 				zap.String("component", "handler.openai_gateway.alpha_search"),

backend/internal/handler/openai_chat_completions.go

@@ -144,11 +144,16 @@ func (h *OpenAIGatewayHandler) ChatCompletions(c *gin.Context) {
 
 	maxAccountSwitches := h.maxAccountSwitches
 	switchCount := 0
+	profitVetoCount := 0
 	failedAccountIDs := make(map[int64]struct{})
 	sameAccountRetryCount := make(map[int64]int)
 	var lastFailoverErr *service.UpstreamFailoverError
 	var oauth429FailoverState service.OpenAIOAuth429FailoverState
 
+	// 分组利润控制：chat completions 文本入口请求级装门并固定 pricingAt。
+	ccPricingCtx, pricingAt := h.gatewayService.WithOpenAIRequestPricingContext(c.Request.Context(), apiKey.GroupID)
+	c.Request = c.Request.WithContext(ccPricingCtx)
+
 	for {
 		if failoverClientGone(c) {
 			return
@@ -207,8 +212,16 @@ func (h *OpenAIGatewayHandler) ChatCompletions(c *gin.Context) {
 		_ = scheduleDecision
 		setOpsSelectedAccount(c, account.ID, account.Platform)
 
-		accountReleaseFunc, acquired := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, reqStream, &streamStarted, reqLog)
-		if !acquired {
+		accountReleaseFunc, slotResult := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, reqStream, &streamStarted, reqLog)
+		if slotResult == openAISlotAcquireProfitVetoed {
+			// 利润终检否决：排除该账号重新选号；否决次数达上限则按无可用账号终止。
+			if !recordOpenAIProfitVeto(failedAccountIDs, account.ID, &profitVetoCount) {
+				h.handleOpenAIProfitVetoExhausted(c, streamStarted, reqLog, profitVetoCount)
+				return
+			}
+			continue
+		}
+		if slotResult != openAISlotAcquireOK {
 			return
 		}
 
@@ -340,6 +353,7 @@ func (h *OpenAIGatewayHandler) ChatCompletions(c *gin.Context) {
 		inboundEndpoint := GetInboundEndpoint(c)
 		upstreamEndpoint := resolveOpenAIUpstreamEndpoint(c, account, result)
 		quotaPlatform := service.QuotaPlatform(c.Request.Context(), apiKey)
+		sessionID := service.ExtractClientSessionID(c)
 
 		cyberBlocked := service.GetOpsCyberPolicy(c) != nil
 		h.submitOpenAIUsageRecordTask(c.Request.Context(), result, func(ctx context.Context) {
@@ -355,7 +369,9 @@ func (h *OpenAIGatewayHandler) ChatCompletions(c *gin.Context) {
 				IPAddress:          clientIP,
 				APIKeyService:      h.apiKeyService,
 				QuotaPlatform:      quotaPlatform,
+				SessionID:          sessionID,
 				ChannelUsageFields: clientRequestedUsageFields(c, channelMapping, reqModel, result.UpstreamModel),
+				PricingAt:          pricingAt,
 				CyberBlocked:       cyberBlocked,
 			}); err != nil {
 				logger.L().With(

backend/internal/handler/openai_codex_models_handler.go

@@ -15,9 +15,9 @@ import (
 // Codex CLI and the Codex desktop app refresh their model picker from
 // GET {base_url}/models?client_version=... (custom provider mode) or
 // GET /backend-api/codex/models (chatgpt_base_url mode). Both routes land
-// here. The manifest is proxied verbatim from the selected account's ChatGPT
-// backend or custom API key upstream. API key manifests use a short-lived,
-// asynchronously revalidated cache to tolerate canceled client requests.
+// here. ChatGPT manifests are proxied verbatim; custom API key manifests receive
+// provider-compatibility normalization and use a short-lived, asynchronously
+// revalidated cache to tolerate canceled client requests.
 func (h *OpenAIGatewayHandler) CodexModels(c *gin.Context) {
 	if c.Request.Context().Err() != nil {
 		return

backend/internal/handler/openai_embeddings.go

@@ -107,6 +107,7 @@ func (h *OpenAIGatewayHandler) Embeddings(c *gin.Context) {
 		return
 	}
 
+	profitVetoCount := 0
 	failedAccountIDs := make(map[int64]struct{})
 	var lastFailoverErr *service.UpstreamFailoverError
 	switchCount := 0
@@ -116,6 +117,10 @@ func (h *OpenAIGatewayHandler) Embeddings(c *gin.Context) {
 	}
 	routingStart := time.Now()
 
+	// 分组利润控制：embeddings 文本入口请求级装门并固定 pricingAt。
+	embPricingCtx, pricingAt := h.gatewayService.WithOpenAIRequestPricingContext(c.Request.Context(), apiKey.GroupID)
+	c.Request = c.Request.WithContext(embPricingCtx)
+
 	for {
 		selection, _, err := h.gatewayService.SelectAccountWithSchedulerForCapability(
 			c.Request.Context(),
@@ -165,8 +170,16 @@ func (h *OpenAIGatewayHandler) Embeddings(c *gin.Context) {
 		account := selection.Account
 		setOpsSelectedAccount(c, account.ID, account.Platform)
 
-		accountReleaseFunc, accountAcquired := h.acquireResponsesAccountSlot(c, apiKey.GroupID, "", selection, false, &streamStarted, reqLog)
-		if !accountAcquired {
+		accountReleaseFunc, slotResult := h.acquireResponsesAccountSlot(c, apiKey.GroupID, "", selection, false, &streamStarted, reqLog)
+		if slotResult == openAISlotAcquireProfitVetoed {
+			// 利润终检否决：排除该账号重新选号；否决次数达上限则按无可用账号终止。
+			if !recordOpenAIProfitVeto(failedAccountIDs, account.ID, &profitVetoCount) {
+				h.handleOpenAIProfitVetoExhausted(c, streamStarted, reqLog, profitVetoCount)
+				return
+			}
+			continue
+		}
+		if slotResult != openAISlotAcquireOK {
 			return
 		}
 
@@ -243,6 +256,7 @@ func (h *OpenAIGatewayHandler) Embeddings(c *gin.Context) {
 		inboundEndpoint := GetInboundEndpoint(c)
 		upstreamEndpoint := GetUpstreamEndpoint(c, account.Platform)
 		quotaPlatform := service.QuotaPlatform(c.Request.Context(), apiKey)
+		sessionID := service.ExtractClientSessionID(c)
 
 		h.submitOpenAIUsageRecordTask(c.Request.Context(), result, func(ctx context.Context) {
 			if err := h.gatewayService.RecordUsage(ctx, &service.OpenAIRecordUsageInput{
@@ -257,7 +271,9 @@ func (h *OpenAIGatewayHandler) Embeddings(c *gin.Context) {
 				IPAddress:          clientIP,
 				APIKeyService:      h.apiKeyService,
 				QuotaPlatform:      quotaPlatform,
+				SessionID:          sessionID,
 				ChannelUsageFields: clientRequestedUsageFields(c, channelMapping, reqModel, result.UpstreamModel),
+				PricingAt:          pricingAt,
 			}); err != nil {
 				logger.L().With(
 					zap.String("component", "handler.openai_gateway.embeddings"),

backend/internal/handler/openai_gateway_count_tokens.go

@@ -143,6 +143,9 @@ func (h *OpenAIGatewayHandler) CountTokens(c *gin.Context) {
 	}
 
 	requestStart := time.Now()
+	// count_tokens 不计费：显式豁免利润门，避免高倍率账号池被门排除后连
+	// token 计数都返回 no available accounts。
+	c.Request = c.Request.WithContext(service.WithOpenAIProfitControlSuppressed(c.Request.Context()))
 	sessionHash := h.gatewayService.GenerateSessionHash(c, body)
 	currentRoutingModel := routingModel
 	if preferredMappedModel != "" {

backend/internal/handler/openai_gateway_credential_failover_test.go

@@ -42,6 +42,28 @@ func TestGatewayChatCredentialStopDoesNotSelectAnotherAccountAndReturnsSafe503(t
 	require.NotContains(t, recorder.Body.String(), "client_secret")
 }
 
+func TestGatewayChatAntigravityCredentialFailureReturnsActionableMessage(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	recorder := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(recorder)
+
+	(&GatewayHandler{}).handleCCFailoverExhausted(c, &service.UpstreamFailoverError{
+		StatusCode:        http.StatusUnauthorized,
+		Stage:             service.GatewayFailureStageAccountAuth,
+		Scope:             service.GatewayFailureScopeAccount,
+		Reason:            service.AntigravityCredentialRejectedReason,
+		NextAccountAction: service.NextAccountRetry,
+		ClientStatusCode:  http.StatusBadGateway,
+		ClientMessage:     service.AntigravityCredentialRejectedClientMessage,
+		ResponseBody:      []byte(`{"error":{"message":"Invalid bearer token","refresh_token":"must-not-leak"}}`),
+	}, false)
+
+	require.Equal(t, http.StatusBadGateway, recorder.Code)
+	require.Contains(t, recorder.Body.String(), service.AntigravityCredentialRejectedClientMessage)
+	require.NotContains(t, strings.ToLower(recorder.Body.String()), "bearer")
+	require.NotContains(t, strings.ToLower(recorder.Body.String()), "refresh_token")
+}
+
 func TestGatewayChatInferenceExhaustionRestoresRetryAfter(t *testing.T) {
 	gin.SetMode(gin.TestMode)
 	recorder := httptest.NewRecorder()

backend/internal/handler/openai_gateway_handler.go

@@ -9,6 +9,8 @@ import (
 	"runtime/debug"
 	"strconv"
 	"strings"
+	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/Wei-Shaw/sub2api/internal/config"
@@ -43,6 +45,49 @@ type OpenAIGatewayHandler struct {
 	cfg                        *config.Config
 }
 
+type openAIWSTurnChannelMappingSnapshot struct {
+	turn    int
+	mapping service.ChannelMappingResult
+}
+
+var errOpenAIWSUnsupportedModelSwitch = errors.New("selected account does not support websocket model switch")
+
+func newOpenAIWSUnsupportedModelSwitchError(model string) error {
+	cause := fmt.Errorf("%w: model %q", errOpenAIWSUnsupportedModelSwitch, strings.TrimSpace(model))
+	return service.NewOpenAIWSClientCloseError(coderws.StatusPolicyViolation, "model switch requires reconnect", cause)
+}
+
+func shouldReportOpenAIWSProxyAccountFailure(err error) bool {
+	return err != nil && !errors.Is(err, errOpenAIWSUnsupportedModelSwitch)
+}
+
+func openAIWSTurnBillingModel(result *service.OpenAIForwardResult, mapping service.ChannelMappingResult, requestedModel, upstreamModel string) string {
+	billingModel := ""
+	if result != nil {
+		billingModel = strings.TrimSpace(result.BillingModel)
+	}
+	if billingModel == "" {
+		billingModel = strings.TrimSpace(upstreamModel)
+	}
+	if billingModel == "" {
+		billingModel = strings.TrimSpace(requestedModel)
+	}
+
+	requestedModel = strings.TrimSpace(requestedModel)
+	switch mapping.BillingModelSource {
+	case service.BillingModelSourceRequested:
+		if requestedModel != "" {
+			billingModel = requestedModel
+		}
+	case service.BillingModelSourceChannelMapped:
+		mappedModel := strings.TrimSpace(mapping.MappedModel)
+		if mappedModel != "" && mappedModel != requestedModel {
+			billingModel = mappedModel
+		}
+	}
+	return billingModel
+}
+
 type grokMediaEligibilityProber interface {
 	ProbeMediaEligibility(ctx context.Context, accountID int64) (bool, string, error)
 }
@@ -382,10 +427,12 @@ func (h *OpenAIGatewayHandler) Responses(c *gin.Context) {
 	maxAccountSwitches := h.maxAccountSwitches
 	switchCount := 0
 	firstOutputTimeoutSwitchCount := 0
+	profitVetoCount := 0
 	failedAccountIDs := make(map[int64]struct{})
 	sameAccountRetryCount := make(map[int64]int)
 	var lastFailoverErr *service.UpstreamFailoverError
 	var oauth429FailoverState service.OpenAIOAuth429FailoverState
+	var passthroughFailoverState openAIPassthroughFailoverState
 
 	// 生图意图的 /v1/responses 请求必须调度到确实支持 Responses API 的账号，否则
 	// 会在 forward 阶段被静默降级为无法生图的 Chat Completions 直转（#4417）。
@@ -394,6 +441,13 @@ func (h *OpenAIGatewayHandler) Responses(c *gin.Context) {
 	// 该判断已排除 Codex 被动 image_gen namespace，避免 CC-only 账号被误过滤（#4476）。
 	requiredCapability := openAIResponsesRequiredCapability(imageIntent, requestPlatform)
 
+	// 分组利润控制：请求级装配定价上下文——pricingAt 固定本请求的
+	// D 与计费高峰因子，选号、槽位终检与全部 failover 重入共用同一门与阈值。
+	// 生图意图只影响能力路由与图片计费，不关门：混合 /v1/responses 请求的
+	// token 计费部分仍受利润门保护，独立图片/视频端点才在门外。
+	pricingCtx, pricingAt := h.gatewayService.WithOpenAIRequestPricingContext(c.Request.Context(), apiKey.GroupID)
+	c.Request = c.Request.WithContext(pricingCtx)
+
 	for {
 		// Streaming Forward intentionally detaches the upstream request so usage can
 		// be drained after a disconnect. Re-check the client context before every
@@ -471,8 +525,17 @@ func (h *OpenAIGatewayHandler) Responses(c *gin.Context) {
 		reqLog.Debug("openai.account_selected", zap.Int64("account_id", account.ID), zap.String("account_name", account.Name))
 		setOpsSelectedAccount(c, account.ID, account.Platform)
 
-		accountReleaseFunc, acquired := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, reqStream, &streamStarted, reqLog)
-		if !acquired {
+		accountReleaseFunc, slotResult := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, reqStream, &streamStarted, reqLog)
+		if slotResult == openAISlotAcquireProfitVetoed {
+			// 利润终检否决：排除该账号重新选号，全池耗尽由下一轮选号报错；
+			// 否决次数达上限则直接终止，避免排队抢槽后才终检的延迟放大。
+			if !recordOpenAIProfitVeto(failedAccountIDs, account.ID, &profitVetoCount) {
+				h.handleOpenAIProfitVetoExhausted(c, streamStarted, reqLog, profitVetoCount)
+				return
+			}
+			continue
+		}
+		if slotResult != openAISlotAcquireOK {
 			return
 		}
 
@@ -482,13 +545,17 @@ func (h *OpenAIGatewayHandler) Responses(c *gin.Context) {
 		// 用扣除 compact 心跳字节的口径快照：心跳注释不构成语义响应，
 		// 不能因心跳字节变化而放弃 failover 换号（#3887）。
 		writerSizeBeforeForward := service.OpenAICompactKeepaliveAdjustedWrittenSize(c)
+		// 跨 passthrough 边界的 failover：从 Kiro 等透传账号切到 Bedrock 等非透传账号前，
+		// 从不可变的 canonical forwardBody 派生本次尝试 body 并整块剔除上游私有的加密
+		// reasoning item（含耦合的 id/summary），避免非透传上游 400 拒绝 Kiro reasoning 形态。
+		attemptBody := h.deriveOpenAIForwardAttemptBody(reqLog, forwardBody, account, &passthroughFailoverState)
 		result, err := func() (*service.OpenAIForwardResult, error) {
 			defer func() {
 				if accountReleaseFunc != nil {
 					accountReleaseFunc()
 				}
 			}()
-			return h.gatewayService.Forward(c.Request.Context(), c, account, forwardBody)
+			return h.gatewayService.Forward(c.Request.Context(), c, account, attemptBody)
 		}()
 		cyberBlockKeyHTTP := ""
 		if service.GetOpsCyberPolicy(c) != nil {
@@ -627,6 +694,7 @@ func (h *OpenAIGatewayHandler) Responses(c *gin.Context) {
 		inboundEndpoint := GetInboundEndpoint(c)
 		upstreamEndpoint := resolveOpenAIUpstreamEndpoint(c, account, result)
 		quotaPlatform := service.QuotaPlatform(c.Request.Context(), apiKey)
+		sessionID := service.ExtractClientSessionID(c)
 
 		// 使用量记录通过有界 worker 池提交，避免请求热路径创建无界 goroutine。
 		cyberBlocked := service.GetOpsCyberPolicy(c) != nil
@@ -644,7 +712,9 @@ func (h *OpenAIGatewayHandler) Responses(c *gin.Context) {
 				RequestPayloadHash: requestPayloadHash,
 				APIKeyService:      h.apiKeyService,
 				QuotaPlatform:      quotaPlatform,
+				SessionID:          sessionID,
 				ChannelUsageFields: clientRequestedUsageFields(c, channelMapping, reqModel, result.UpstreamModel),
+				PricingAt:          pricingAt,
 				CyberBlocked:       cyberBlocked,
 			}); err != nil {
 				logger.L().With(
@@ -937,12 +1007,17 @@ func (h *OpenAIGatewayHandler) Messages(c *gin.Context) {
 
 	maxAccountSwitches := h.maxAccountSwitches
 	switchCount := 0
+	profitVetoCount := 0
 	failedAccountIDs := make(map[int64]struct{})
 	sameAccountRetryCount := make(map[int64]int)
 	var lastFailoverErr *service.UpstreamFailoverError
 	var oauth429FailoverState service.OpenAIOAuth429FailoverState
 	effectiveMappedModel := preferredMappedModel
 
+	// 分组利润控制：Messages 文本入口同样请求级装门并固定 pricingAt。
+	msgPricingCtx, pricingAt := h.gatewayService.WithOpenAIRequestPricingContext(c.Request.Context(), apiKey.GroupID)
+	c.Request = c.Request.WithContext(msgPricingCtx)
+
 	for {
 		if failoverClientGone(c) {
 			return
@@ -1007,8 +1082,17 @@ func (h *OpenAIGatewayHandler) Messages(c *gin.Context) {
 		_ = scheduleDecision
 		setOpsSelectedAccount(c, account.ID, account.Platform)
 
-		accountReleaseFunc, acquired := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, reqStream, &streamStarted, reqLog)
-		if !acquired {
+		accountReleaseFunc, slotResult := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, reqStream, &streamStarted, reqLog)
+		if slotResult == openAISlotAcquireProfitVetoed {
+			// 利润终检否决：排除该账号重新选号，全池耗尽由下一轮选号报错；
+			// 否决次数达上限则直接终止，避免排队抢槽后才终检的延迟放大。
+			if !recordOpenAIProfitVeto(failedAccountIDs, account.ID, &profitVetoCount) {
+				h.handleOpenAIProfitVetoExhausted(c, streamStarted, reqLog, profitVetoCount)
+				return
+			}
+			continue
+		}
+		if slotResult != openAISlotAcquireOK {
 			return
 		}
 
@@ -1138,6 +1222,7 @@ func (h *OpenAIGatewayHandler) Messages(c *gin.Context) {
 		inboundEndpoint := GetInboundEndpoint(c)
 		upstreamEndpoint := resolveOpenAIUpstreamEndpoint(c, account, result)
 		quotaPlatform := service.QuotaPlatform(c.Request.Context(), apiKey)
+		sessionID := service.ExtractClientSessionID(c)
 
 		cyberBlocked := service.GetOpsCyberPolicy(c) != nil
 		h.submitOpenAIUsageRecordTask(c.Request.Context(), result, func(ctx context.Context) {
@@ -1154,7 +1239,9 @@ func (h *OpenAIGatewayHandler) Messages(c *gin.Context) {
 				RequestPayloadHash: requestPayloadHash,
 				APIKeyService:      h.apiKeyService,
 				QuotaPlatform:      quotaPlatform,
+				SessionID:          sessionID,
 				ChannelUsageFields: clientRequestedUsageFields(c, channelMappingMsg, reqModel, result.UpstreamModel),
+				PricingAt:          pricingAt,
 				CyberBlocked:       cyberBlocked,
 			}); err != nil {
 				logger.L().With(
@@ -1295,6 +1382,74 @@ func (h *OpenAIGatewayHandler) acquireResponsesUserSlot(
 	return wrapReleaseOnDone(ctx, userReleaseFunc), true
 }
 
+// openAISlotAcquireResult 是账号槽位获取的三态结果。
+type openAISlotAcquireResult int
+
+const (
+	openAISlotAcquireOK openAISlotAcquireResult = iota
+	// openAISlotAcquireFailed：错误响应已写出，调用方直接 return。
+	openAISlotAcquireFailed
+	// openAISlotAcquireProfitVetoed：槽位获取成功后利润终检否决。槽位已释放、
+	// 未写任何响应；调用方应经 recordOpenAIProfitVeto 把该账号加入本请求排除集
+	// 后重新选号，全池耗尽由下一轮选号返回标准 no available accounts。
+	openAISlotAcquireProfitVetoed
+)
+
+// openAIWSTurnPricing 持有 WebSocket 连接内「当前 turn」的计费定价时刻。
+// 由 BeforeTurn 在每个 turn 开始时冻结，AfterTurn 的用量提交读取它；turn 在
+// 连接内串行推进，互斥锁只为跨用量提交 goroutine 的读取安全。
+//
+// 零值语义（重要）：ws_v2 passthrough ingress 只实现了 AfterTurn，没有任何
+// turn 起始回调，BeforeTurn 永远不会被调用。此时本值保持零，RecordUsage 经
+// openAIUsagePricingAt 回退到记录时刻——与引入分组利润控制前的基线一致。
+// 绝不能用建连时刻初始化：那会把透传连接的所有 turn 钉死在建连时的高峰因子，
+// 客户端只要峰前一分钟建连并保活，整条连接就能全程按谷价结算，正是利润控制
+// 想堵的漏洞。透传 ingress 目前不做 turn 级利润复核，只有建连时的准入门。
+type openAIWSTurnPricing struct {
+	mu sync.Mutex
+	at time.Time
+}
+
+func (p *openAIWSTurnPricing) freeze(at time.Time) {
+	p.mu.Lock()
+	p.at = at
+	p.mu.Unlock()
+}
+
+func (p *openAIWSTurnPricing) current() time.Time {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	return p.at
+}
+
+// recordOpenAIProfitVeto 记录 OpenAI 侧选号循环的一次利润门终检否决：把账号
+// 加入本请求排除集并递增否决计数。返回 false 表示否决次数已达
+// maxProfitVetoAttempts，调用方必须停止重选并按「无可用账号」终止。
+//
+// OpenAI 路径用的是各自的 failedAccountIDs map + for 循环（不是 FailoverState），
+// 这里用一个独立计数器复用同一上限语义。上限是必需的：WaitPlan 分支先阻塞
+// 排队（sticky 45s / fallback 30s）拿到槽位才终检，无上限重选会把单次请求的
+// 延迟放大到 N × WaitPlan.Timeout。
+func recordOpenAIProfitVeto(failedAccountIDs map[int64]struct{}, accountID int64, vetoCount *int) bool {
+	failedAccountIDs[accountID] = struct{}{}
+	*vetoCount++
+	return *vetoCount < maxProfitVetoAttempts
+}
+
+// handleOpenAIProfitVetoExhausted 在利润否决预算耗尽时写出错误响应。
+// 与 acquireResponsesAccountSlot 内部的 no-available-accounts 失败分支同形，
+// 保证同一调用方在两条路径上拿到一致的响应格式。
+func (h *OpenAIGatewayHandler) handleOpenAIProfitVetoExhausted(
+	c *gin.Context,
+	streamStarted bool,
+	reqLog *zap.Logger,
+	vetoCount int,
+) {
+	reqLog.Warn("openai.profit_veto_attempts_exhausted", zap.Int("profit_veto_count", vetoCount))
+	markOpsRoutingCapacityLimited(c)
+	h.handleStreamingAwareError(c, http.StatusServiceUnavailable, "api_error", profitVetoExhaustedMessage, streamStarted)
+}
+
 func (h *OpenAIGatewayHandler) acquireResponsesAccountSlot(
 	c *gin.Context,
 	groupID *int64,
@@ -1303,22 +1458,41 @@ func (h *OpenAIGatewayHandler) acquireResponsesAccountSlot(
 	reqStream bool,
 	streamStarted *bool,
 	reqLog *zap.Logger,
-) (func(), bool) {
+) (func(), openAISlotAcquireResult) {
 	if selection == nil || selection.Account == nil {
 		markOpsRoutingCapacityLimited(c)
 		h.handleStreamingAwareError(c, http.StatusServiceUnavailable, "api_error", "No available accounts", *streamStarted)
-		return nil, false
+		return nil, openAISlotAcquireFailed
 	}
 
-	ctx := c.Request.Context()
+	// 终检与准入后绑定使用选号结果携带的门：composite 等跨分组调度解析出的
+	// 门只存在于调度栈的局部 ctx，必须经选号结果重放到本函数的 ctx 上。
+	ctx := service.ContextWithSelectionProfitGate(c.Request.Context(), selection)
 	account := selection.Account
 	if selection.Acquired {
-		return wrapReleaseOnDone(ctx, selection.ReleaseFunc), true
+		latest, vetoed, reason := h.gatewayService.ProfitControlVetoLatest(ctx, account)
+		if vetoed {
+			if selection.ReleaseFunc != nil {
+				selection.ReleaseFunc()
+			}
+			reqLog.Debug("openai.account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+			return nil, openAISlotAcquireProfitVetoed
+		}
+		account = latest
+		selection.Account = latest
+		// 调度器已抢槽路径无门时由选号内部完成 eager 绑定；门下选号内部
+		// 推迟绑定，这里在终检通过后补准入后绑定。
+		if selection.ProfitGateActive() {
+			if err := h.gatewayService.BindStickySessionAfterProfitAdmission(ctx, groupID, sessionHash, account.ID); err != nil {
+				reqLog.Warn("openai.bind_sticky_session_after_profit_admission_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+			}
+		}
+		return wrapReleaseOnDone(ctx, selection.ReleaseFunc), openAISlotAcquireOK
 	}
 	if selection.WaitPlan == nil {
 		markOpsRoutingCapacityLimited(c)
 		h.handleStreamingAwareError(c, http.StatusServiceUnavailable, "api_error", "No available accounts", *streamStarted)
-		return nil, false
+		return nil, openAISlotAcquireFailed
 	}
 
 	fastReleaseFunc, fastAcquired, err := h.concurrencyHelper.TryAcquireAccountSlot(
@@ -1329,13 +1503,25 @@ func (h *OpenAIGatewayHandler) acquireResponsesAccountSlot(
 	if err != nil {
 		reqLog.Warn("openai.account_slot_quick_acquire_failed", zap.Int64("account_id", account.ID), zap.Error(err))
 		h.handleConcurrencyError(c, err, "account", *streamStarted)
-		return nil, false
+		return nil, openAISlotAcquireFailed
 	}
 	if fastAcquired {
-		if err := h.gatewayService.BindStickySession(ctx, groupID, sessionHash, account.ID); err != nil {
-			reqLog.Warn("openai.bind_sticky_session_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+		// 分组利润控制：快速抢槽成功后终检。选号与抢槽之间账号
+		// 倍率可能刷新，越线则释放槽位交由调用方排除重选，不绑定粘连。
+		latest, vetoed, reason := h.gatewayService.ProfitControlVetoLatest(ctx, account)
+		if vetoed {
+			if fastReleaseFunc != nil {
+				fastReleaseFunc()
+			}
+			reqLog.Debug("openai.account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+			return nil, openAISlotAcquireProfitVetoed
 		}
-		return wrapReleaseOnDone(ctx, fastReleaseFunc), true
+		account = latest
+		selection.Account = latest
+		if err := h.gatewayService.BindStickySessionAfterProfitAdmission(ctx, groupID, sessionHash, account.ID); err != nil {
+			reqLog.Warn("openai.bind_sticky_session_after_profit_admission_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+		}
+		return wrapReleaseOnDone(ctx, fastReleaseFunc), openAISlotAcquireOK
 	}
 
 	canWait, waitErr := h.concurrencyHelper.IncrementAccountWaitCount(ctx, account.ID, selection.WaitPlan.MaxWaiting)
@@ -1347,7 +1533,7 @@ func (h *OpenAIGatewayHandler) acquireResponsesAccountSlot(
 			zap.Int("max_waiting", selection.WaitPlan.MaxWaiting),
 		)
 		h.handleStreamingAwareError(c, http.StatusTooManyRequests, "rate_limit_error", "Too many pending requests, please retry later", *streamStarted)
-		return nil, false
+		return nil, openAISlotAcquireFailed
 	}
 
 	accountWaitCounted := waitErr == nil && canWait
@@ -1370,15 +1556,27 @@ func (h *OpenAIGatewayHandler) acquireResponsesAccountSlot(
 	if err != nil {
 		reqLog.Warn("openai.account_slot_acquire_failed", zap.Int64("account_id", account.ID), zap.Error(err))
 		h.handleConcurrencyError(c, err, "account", *streamStarted)
-		return nil, false
+		return nil, openAISlotAcquireFailed
 	}
 
 	// Slot acquired: no longer waiting in queue.
 	releaseWait()
-	if err := h.gatewayService.BindStickySession(ctx, groupID, sessionHash, account.ID); err != nil {
-		reqLog.Warn("openai.bind_sticky_session_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+	// 分组利润控制：WaitPlan 排队成功后终检。排队期间账号倍率
+	// 可能上调，越线则释放槽位交由调用方排除重选，不绑定粘连。
+	latest, vetoed, reason := h.gatewayService.ProfitControlVetoLatest(ctx, account)
+	if vetoed {
+		if accountReleaseFunc != nil {
+			accountReleaseFunc()
+		}
+		reqLog.Debug("openai.account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+		return nil, openAISlotAcquireProfitVetoed
 	}
-	return wrapReleaseOnDone(ctx, accountReleaseFunc), true
+	account = latest
+	selection.Account = latest
+	if err := h.gatewayService.BindStickySessionAfterProfitAdmission(ctx, groupID, sessionHash, account.ID); err != nil {
+		reqLog.Warn("openai.bind_sticky_session_after_profit_admission_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+	}
+	return wrapReleaseOnDone(ctx, accountReleaseFunc), openAISlotAcquireOK
 }
 
 // ResponsesWebSocket handles OpenAI Responses API WebSocket ingress endpoint
@@ -1515,7 +1713,7 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 	previousResponseCanMove := !firstMessageToolCoverage.HasFunctionCallOutput || firstMessageToolCoverage.ContextCoversAllCallIDs
 	reqLog = reqLog.With(
 		zap.Bool("ws_ingress", true),
-		zap.String("model", reqModel),
+		zap.String("session_initial_model", reqModel),
 		zap.Bool("has_previous_response_id", previousResponseID != ""),
 		zap.String("previous_response_id_kind", previousResponseIDKind),
 	)
@@ -1614,6 +1812,7 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 	)
 	maxAccountSwitches := h.maxAccountSwitches
 	switchCount := 0
+	profitVetoCount := 0
 	failedAccountIDs := make(map[int64]struct{})
 	var lastFailoverErr *service.UpstreamFailoverError
 	var oauth429FailoverState service.OpenAIOAuth429FailoverState
@@ -1664,6 +1863,14 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 		requiredCapability = service.OpenAIEndpointCapabilityResponses
 	}
 
+	// 分组利润控制：WS 桥按连接装配定价上下文并装门（选号与抢槽共用该
+	// ctx）。连接内不重选号，但每个 turn 开始经 BeforeTurn 重新冻结 pricingAt
+	// 并按最新门复核当前账号（准入与计费同源），峰前建连保活不能让后续 turn
+	// 继续按建连时刻的谷价计费。生图意图只影响能力路由与图片计费，不关门。
+	// 建连时刻只用于选号/准入，不作为任何 turn 的计费定价时刻。
+	wsPricingCtx, _ := h.gatewayService.WithOpenAIRequestPricingContext(ctx, apiKey.GroupID)
+	ctx = wsPricingCtx
+
 	for {
 		if ctx.Err() != nil {
 			return
@@ -1709,7 +1916,28 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 		if selection.WaitPlan != nil && selection.WaitPlan.MaxConcurrency > 0 {
 			accountMaxConcurrency = selection.WaitPlan.MaxConcurrency
 		}
+		// 终检、准入后绑定与后续 turn 级复核都使用选号结果携带的门（composite
+		// 等跨分组调度的门只存在于调度栈局部 ctx）；准入成功后并入连接 ctx。
+		admissionCtx := service.ContextWithSelectionProfitGate(ctx, selection)
 		accountReleaseFunc := selection.ReleaseFunc
+		if selection.Acquired {
+			// 调度器已抢槽路径同样终检：选号与抢槽之间账号倍率可能刷新。
+			latest, vetoed, reason := h.gatewayService.ProfitControlVetoLatest(admissionCtx, account)
+			if vetoed {
+				if accountReleaseFunc != nil {
+					accountReleaseFunc()
+				}
+				reqLog.Debug("openai.websocket_account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+				if !recordOpenAIProfitVeto(failedAccountIDs, account.ID, &profitVetoCount) {
+					reqLog.Warn("openai.websocket_profit_veto_attempts_exhausted", zap.Int("profit_veto_count", profitVetoCount))
+					closeOpenAIClientWS(wsConn, coderws.StatusTryAgainLater, "no available account")
+					return
+				}
+				continue
+			}
+			account = latest
+			selection.Account = latest
+		}
 		if !selection.Acquired {
 			if selection.WaitPlan == nil {
 				closeOpenAIClientWS(wsConn, coderws.StatusTryAgainLater, "account is busy, please retry later")
@@ -1729,11 +1957,30 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 				closeOpenAIClientWS(wsConn, coderws.StatusTryAgainLater, "account is busy, please retry later")
 				return
 			}
+			// 分组利润控制：WS 快速抢槽成功后终检，越线则释放
+			// 槽位、排除该账号重新选号，全池耗尽由下一轮选号关闭连接。
+			latest, vetoed, reason := h.gatewayService.ProfitControlVetoLatest(admissionCtx, account)
+			if vetoed {
+				if fastReleaseFunc != nil {
+					fastReleaseFunc()
+				}
+				reqLog.Debug("openai.websocket_account_slot_profit_vetoed", zap.Int64("account_id", account.ID), zap.String("reason", reason))
+				if !recordOpenAIProfitVeto(failedAccountIDs, account.ID, &profitVetoCount) {
+					reqLog.Warn("openai.websocket_profit_veto_attempts_exhausted", zap.Int("profit_veto_count", profitVetoCount))
+					closeOpenAIClientWS(wsConn, coderws.StatusTryAgainLater, "no available account")
+					return
+				}
+				continue
+			}
+			account = latest
+			selection.Account = latest
 			accountReleaseFunc = fastReleaseFunc
 		}
+		// 准入完成：门并入连接 ctx，turn 级复核与 failover 重选共用。
+		ctx = admissionCtx
 		currentAccountRelease = wrapReleaseOnDone(ctx, accountReleaseFunc)
-		if err := h.gatewayService.BindStickySession(ctx, apiKey.GroupID, sessionHash, account.ID); err != nil {
-			reqLog.Warn("openai.websocket_bind_sticky_session_failed", zap.Int64("account_id", account.ID), zap.Error(err))
+		if err := h.gatewayService.BindStickySessionAfterProfitAdmission(ctx, apiKey.GroupID, sessionHash, account.ID); err != nil {
+			reqLog.Warn("openai.websocket_bind_sticky_session_after_profit_admission_failed", zap.Int64("account_id", account.ID), zap.Error(err))
 		}
 
 		token, _, err := h.gatewayService.GetRequestCredential(ctx, c, account)
@@ -1767,6 +2014,14 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 			reasoningEffortMappings = apiKey.Group.ReasoningEffortMappings
 		}
 		var requestPayloadHash string
+		// Passthrough rejects overlapping response.create frames, so one immutable
+		// turn-tagged slot preserves the exact mapping used for the in-flight request.
+		var turnChannelMapping atomic.Pointer[openAIWSTurnChannelMappingSnapshot]
+		turnChannelMapping.Store(&openAIWSTurnChannelMappingSnapshot{turn: 1, mapping: channelMappingWS})
+		// turn 级定价：BeforeTurn 重新冻结 pricingAt 并按最新门复核当前账号，
+		// AfterTurn 的计费读取所属 turn 的时刻。零值起步的语义见
+		// openAIWSTurnPricing 的注释——绝不能用建连时刻初始化。
+		var turnPricing openAIWSTurnPricing
 		hooks := &service.OpenAIWSIngressHooks{
 			InitialRequestModel:     reqModel,
 			MaxReasoningEffort:      maxReasoningEffort,
@@ -1791,11 +2046,39 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 				}
 				return nil
 			},
+			MapRequestModel: func(turn int, originalModel string) (string, error) {
+				model := strings.TrimSpace(originalModel)
+				if model == "" {
+					model = reqModel
+				}
+				mapping, _ := h.gatewayService.ResolveChannelMappingAndRestrict(ctx, apiKey.GroupID, model)
+				mappedModelUnchanged := false
+				if previous := turnChannelMapping.Load(); previous != nil && previous.turn < turn {
+					mappedModelUnchanged = strings.TrimSpace(previous.mapping.MappedModel) == strings.TrimSpace(mapping.MappedModel)
+				}
+				if turn > 1 && !mappedModelUnchanged && !account.IsModelSupported(model) && !account.IsModelSupported(mapping.MappedModel) {
+					return "", newOpenAIWSUnsupportedModelSwitchError(mapping.MappedModel)
+				}
+				turnChannelMapping.Store(&openAIWSTurnChannelMappingSnapshot{turn: turn, mapping: mapping})
+				return mapping.MappedModel, nil
+			},
 			BeforeTurn: func(turn int) error {
 				// turn==1 的会话屏蔽已由握手层检查覆盖；连接内 flag 只拦截后续 turn。
 				if cyberBlockedThisConn {
 					return service.NewOpenAIWSClientCloseError(coderws.StatusPolicyViolation, cyberSessionBlockedClientMsg, nil)
 				}
+				// 长连接跨峰谷/倍率刷新防护：每个 turn 按当前时刻重装门并复核
+				// 当前账号，越线即要求客户端重连重选（连接绑定单一上游账号，
+				// 无法中途换号）。本 turn 的准入与计费共用同一 pricingAt。
+				turnCtx, turnAt := h.gatewayService.WithOpenAITurnPricingContext(ctx, apiKey.GroupID)
+				if _, vetoed, reason := h.gatewayService.ProfitControlVetoLatest(turnCtx, account); vetoed {
+					reqLog.Info("openai.websocket_turn_profit_vetoed",
+						zap.Int("turn", turn),
+						zap.Int64("account_id", account.ID),
+						zap.String("reason", reason))
+					return service.NewOpenAIWSClientCloseError(coderws.StatusTryAgainLater, "account is no longer eligible for this connection, please reconnect", nil)
+				}
+				turnPricing.freeze(turnAt)
 				if turn == 1 {
 					return nil
 				}
@@ -1832,7 +2115,27 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 				// 届时 defer 已清除标记）。
 				defer clearCyberPolicyTurnState(c)
 				releaseTurnSlots()
-				h.recordCyberPolicyIfMarked(c, apiKey, account, subscription, reqModel, turnErr != nil, cyberBlockKey, clientRequestedUsageFields(c, channelMappingWS, reqModel, ""), requestPayloadHash)
+				turnRequestedModel := reqModel
+				turnUpstreamModel := ""
+				if result != nil && turn > 1 {
+					if model := strings.TrimSpace(result.Model); model != "" {
+						turnRequestedModel = model
+					}
+				}
+				if result != nil {
+					turnUpstreamModel = strings.TrimSpace(result.UpstreamModel)
+				}
+				var turnMapping service.ChannelMappingResult
+				if snapshot := turnChannelMapping.Load(); snapshot != nil && snapshot.turn == turn {
+					turnMapping = snapshot.mapping
+				} else {
+					turnMapping, _ = h.gatewayService.ResolveChannelMappingAndRestrict(ctx, apiKey.GroupID, turnRequestedModel)
+				}
+				if turnUpstreamModel == "" {
+					turnUpstreamModel = turnRequestedModel
+				}
+				turnUsageFields := turnMapping.ToUsageFields(turnRequestedModel, turnUpstreamModel)
+				h.recordCyberPolicyIfMarked(c, apiKey, account, subscription, turnRequestedModel, turnErr != nil, cyberBlockKey, turnUsageFields, requestPayloadHash)
 				if service.GetOpsCyberPolicy(c) != nil {
 					cyberBlockedThisConn = true
 				}
@@ -1854,14 +2157,27 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 				if result == nil {
 					return
 				}
+				result.BillingModel = openAIWSTurnBillingModel(result, turnMapping, turnRequestedModel, turnUpstreamModel)
+				reqLog.Debug("openai.websocket_turn_billing",
+					zap.Int("turn", turn),
+					zap.String("turn_requested_model", turnRequestedModel),
+					zap.String("turn_upstream_model", turnUpstreamModel),
+					zap.String("billing_model", result.BillingModel),
+				)
 				// 排除 spark 影子:其 codex_* 仅由 QueryUsage(/wham/usage bengalfox)更新(外审第7轮 P1)。
 				if account.Type == service.AccountTypeOAuth && !account.IsShadow() {
 					h.gatewayService.UpdateCodexUsageSnapshotFromHeaders(ctx, account.ID, result.ResponseHeaders)
 				}
-				h.gatewayService.ReportOpenAIAccountScheduleResult(account.ID, account.GetMappedModel(reqModel), openAIForwardSucceededForScheduling(result), result.FirstTokenMs)
+				scheduleModel := turnUpstreamModel
+				if scheduleModel == "" {
+					scheduleModel = turnRequestedModel
+				}
+				h.gatewayService.ReportOpenAIAccountScheduleResult(account.ID, scheduleModel, openAIForwardSucceededForScheduling(result), result.FirstTokenMs)
 				inboundEndpoint := GetInboundEndpoint(c)
 				upstreamEndpoint := resolveOpenAIUpstreamEndpoint(c, account, result)
 				quotaPlatform := service.QuotaPlatform(c.Request.Context(), apiKey)
+				sessionID := service.ExtractClientSessionID(c)
+				turnRecordPricingAt := turnPricing.current()
 				cyberBlocked := service.GetOpsCyberPolicy(c) != nil
 				h.submitOpenAIUsageRecordTask(ctx, result, func(taskCtx context.Context) {
 					if err := h.gatewayService.RecordUsage(taskCtx, &service.OpenAIRecordUsageInput{
@@ -1877,7 +2193,9 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 						RequestPayloadHash: requestPayloadHash,
 						APIKeyService:      h.apiKeyService,
 						QuotaPlatform:      quotaPlatform,
-						ChannelUsageFields: clientRequestedUsageFields(c, channelMappingWS, reqModel, result.UpstreamModel),
+						SessionID:          sessionID,
+						ChannelUsageFields: turnUsageFields,
+						PricingAt:          turnRecordPricingAt,
 						CyberBlocked:       cyberBlocked,
 					}); err != nil {
 						reqLog.Error("openai.websocket_record_usage_failed",
@@ -1890,11 +2208,7 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 			},
 		}
 
-		// 应用渠道模型映射到 WebSocket 首条消息
 		wsFirstMessage := firstMessage
-		if channelMappingWS.Mapped {
-			wsFirstMessage = h.gatewayService.ReplaceModelInBody(firstMessage, channelMappingWS.MappedModel)
-		}
 		// 切组/会话失配防护：previous_response_id 未在当前分组命中粘连账号（StickyPreviousHit=false），
 		// 说明该会话链不属于本次调度到的账号，原样转发会触发上游会话链鉴权失败（“鉴权失败，请检查 API Key”）。
 		// 故剥离首包里的 previous_response_id，改用首包内 input 重建上下文；带 function_call_output 的
@@ -1938,7 +2252,9 @@ func (h *OpenAIGatewayHandler) ResponsesWebSocket(c *gin.Context) {
 				return
 			}
 
-			h.gatewayService.ReportOpenAIAccountScheduleResult(account.ID, account.GetMappedModel(reqModel), false, nil)
+			if shouldReportOpenAIWSProxyAccountFailure(err) {
+				h.gatewayService.ReportOpenAIAccountScheduleResult(account.ID, account.GetMappedModel(reqModel), false, nil)
+			}
 			closeStatus, closeReason := summarizeWSCloseErrorForLog(err)
 			proxyFailedFields := []zap.Field{
 				zap.Int64("account_id", account.ID),
@@ -2079,10 +2395,16 @@ func (h *OpenAIGatewayHandler) submitUsageRecordTask(parent context.Context, tas
 	}
 	task = wrapUsageRecordTaskContext(parent, task)
 	if h.usageRecordWorkerPool != nil {
-		h.usageRecordWorkerPool.Submit(task)
-		return
+		if mode := h.usageRecordWorkerPool.Submit(task); mode != service.UsageRecordSubmitModeDroppedStopped {
+			return
+		}
+		// 池已停止（进程关停窗口）：计费任务不能静默丢失，降级为内联同步执行。
+		// 显式配置的 drop/sample 溢出丢弃仍按配置语义保留。
+		logger.L().With(
+			zap.String("component", "handler.openai_gateway.responses"),
+		).Warn("openai.usage_record_task_stopped_sync_fallback")
 	}
-	// 回退路径：worker 池未注入时同步执行，避免退回到无界 goroutine 模式。
+	// 回退路径：worker 池未注入或已停止时同步执行，避免退回到无界 goroutine 模式。
 	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
 	defer cancel()
 	defer func() {
@@ -2110,7 +2432,7 @@ func (h *OpenAIGatewayHandler) submitMandatoryUsageRecordTask(parent context.Con
 	}
 	task = wrapUsageRecordTaskContext(parent, task)
 	if h.usageRecordWorkerPool != nil {
-		if mode := h.usageRecordWorkerPool.Submit(task); mode != service.UsageRecordSubmitModeDropped {
+		if mode := h.usageRecordWorkerPool.Submit(task); !mode.Dropped() {
 			return
 		}
 		logger.L().With(
@@ -2221,7 +2543,9 @@ func (h *OpenAIGatewayHandler) handleFailoverExhausted(c *gin.Context, failoverE
 }
 
 func credentialFailoverClientResponse(failoverErr *service.UpstreamFailoverError) (int, string) {
-	_ = failoverErr
+	if failoverErr != nil && failoverErr.Reason == service.AntigravityCredentialRejectedReason {
+		return http.StatusBadGateway, service.AntigravityCredentialRejectedClientMessage
+	}
 	return http.StatusServiceUnavailable, service.GrokCredentialUnavailableClientMessage
 }
 
@@ -2884,6 +3208,8 @@ func (h *OpenAIGatewayHandler) recordCyberPolicyIfMarked(c *gin.Context, apiKey
 		userAgent = c.GetHeader("User-Agent")
 		clientIPStr = strings.TrimSpace(ip.GetClientIP(c))
 	}
+	// 提前拍成标量，避免在下方 goroutine 内访问 gin.Context。
+	sessionID := service.ExtractClientSessionID(c)
 	apiKeyPrefix := ""
 	if apiKey != nil {
 		apiKeyPrefix = keyPrefix(apiKey.Key, 8)
@@ -2940,6 +3266,7 @@ func (h *OpenAIGatewayHandler) recordCyberPolicyIfMarked(c *gin.Context, apiKey
 				UpstreamEndpoint:   upstreamEndpoint,
 				UserAgent:          userAgent,
 				IPAddress:          clientIPStr,
+				SessionID:          sessionID,
 				RequestPayloadHash: requestPayloadHash,
 				APIKeyService:      apiKeySvc,
 				ChannelUsageFields: channelFields,

backend/internal/handler/openai_gateway_handler_test.go

@@ -4,6 +4,7 @@ import (
 	"context"
 	"encoding/json"
 	"errors"
+	"fmt"
 	"io"
 	"net/http"
 	"net/http/httptest"
@@ -1265,6 +1266,7 @@ func TestOpenAIResponsesWebSocket_ContentModerationBlocksFirstFrame(t *testing.T
 		nil,
 		nil,
 		nil,
+		nil,
 	)
 	decision, err := moderationSvc.Check(context.Background(), service.ContentModerationCheckInput{
 		UserID:   1,
@@ -1369,6 +1371,229 @@ func TestOpenAIResponsesWebSocket_PassthroughUsageLogLeavesUserAgentNilWhenMissi
 	require.Equal(t, "medium", *got.log.ReasoningEffort)
 }
 
+func TestOpenAIResponsesWebSocket_PassthroughTracksModelPerTurn(t *testing.T) {
+	got := runOpenAIResponsesWebSocketUsageLogCase(t, openAIResponsesWSUsageLogCase{
+		firstPayload:  `{"type":"response.create","model":"sol","stream":false}`,
+		secondPayload: `{"type":"response.create","model":"terra","stream":false}`,
+		channelMapping: map[string]string{
+			"sol":   "sol-channel",
+			"terra": "terra-channel",
+		},
+		accountModelMapping: map[string]any{
+			"sol":           "gpt-5.6-sol",
+			"terra":         "gpt-5.6-terra",
+			"sol-channel":   "gpt-5.6-sol",
+			"terra-channel": "gpt-5.6-terra",
+		},
+	})
+
+	require.Len(t, got.upstreamPayloads, 2)
+	require.Equal(t, "gpt-5.6-sol", gjson.GetBytes(got.upstreamPayloads[0], "model").String())
+	require.Equal(t, "gpt-5.6-terra", gjson.GetBytes(got.upstreamPayloads[1], "model").String())
+	require.Len(t, got.clientEvents, 2)
+	require.Equal(t, "sol", gjson.GetBytes(got.clientEvents[0], "response.model").String())
+	require.Equal(t, "terra", gjson.GetBytes(got.clientEvents[1], "response.model").String())
+
+	require.Len(t, got.logs, 2)
+	require.Equal(t, "sol", got.logs[0].Model)
+	require.Equal(t, "sol", got.logs[0].RequestedModel)
+	require.NotNil(t, got.logs[0].UpstreamModel)
+	require.Equal(t, "gpt-5.6-sol", *got.logs[0].UpstreamModel)
+	require.NotNil(t, got.logs[0].ModelMappingChain)
+	require.Equal(t, "sol→sol-channel→gpt-5.6-sol", *got.logs[0].ModelMappingChain)
+
+	require.Equal(t, "terra", got.logs[1].Model)
+	require.Equal(t, "terra", got.logs[1].RequestedModel)
+	require.NotNil(t, got.logs[1].UpstreamModel)
+	require.Equal(t, "gpt-5.6-terra", *got.logs[1].UpstreamModel)
+	require.NotNil(t, got.logs[1].ModelMappingChain)
+	require.Equal(t, "terra→terra-channel→gpt-5.6-terra", *got.logs[1].ModelMappingChain)
+	require.InDelta(t, got.logs[1].TotalCost*2.5, got.logs[0].TotalCost, 1e-12,
+		"each turn must be billed with its own channel-mapped model")
+}
+
+func TestOpenAIResponsesWebSocket_UnchangedChannelTargetOutsideAccountMappingKeysRemainsValid(t *testing.T) {
+	got := runOpenAIResponsesWebSocketUsageLogCase(t, openAIResponsesWSUsageLogCase{
+		firstPayload:  `{"type":"response.create","model":"public-alias","stream":false}`,
+		secondPayload: `{"type":"response.create","stream":false}`,
+		channelMapping: map[string]string{
+			"public-alias": "gpt-5.6-sol",
+		},
+		accountModelMapping: map[string]any{
+			"public-alias": "gpt-5.6-terra",
+		},
+	})
+
+	require.Len(t, got.upstreamPayloads, 2)
+	require.Equal(t, "gpt-5.6-sol", gjson.GetBytes(got.upstreamPayloads[0], "model").String())
+	require.Equal(t, "gpt-5.6-sol", gjson.GetBytes(got.upstreamPayloads[1], "model").String())
+	require.Len(t, got.clientEvents, 2)
+	require.Equal(t, "public-alias", gjson.GetBytes(got.clientEvents[0], "response.model").String())
+	require.Equal(t, "public-alias", gjson.GetBytes(got.clientEvents[1], "response.model").String())
+	require.Len(t, got.logs, 2)
+	for _, usageLog := range got.logs {
+		require.Equal(t, "public-alias", usageLog.RequestedModel)
+		require.NotNil(t, usageLog.UpstreamModel)
+		require.Equal(t, "gpt-5.6-sol", *usageLog.UpstreamModel)
+		require.NotNil(t, usageLog.ModelMappingChain)
+		require.Equal(t, "public-alias→gpt-5.6-sol", *usageLog.ModelMappingChain)
+	}
+}
+
+func TestOpenAIResponsesWebSocket_PassthroughKeepsTurnMappingSnapshot(t *testing.T) {
+	got := runOpenAIResponsesWebSocketUsageLogCase(t, openAIResponsesWSUsageLogCase{
+		firstPayload:  `{"type":"response.create","model":"sol","stream":false}`,
+		secondPayload: `{"type":"response.create","model":"sol","stream":false}`,
+		channelMapping: map[string]string{
+			"sol": "gpt-5.6-sol",
+		},
+		afterFirstUpstreamRequest: func(channelSvc *service.ChannelService) error {
+			if channelSvc == nil {
+				return errors.New("channel service is nil")
+			}
+			_, err := channelSvc.Update(context.Background(), 7701, &service.UpdateChannelInput{
+				ModelMapping: map[string]map[string]string{
+					service.PlatformOpenAI: {"sol": "gpt-5.6-terra"},
+				},
+			})
+			return err
+		},
+	})
+
+	require.Len(t, got.upstreamPayloads, 2)
+	require.Equal(t, "gpt-5.6-sol", gjson.GetBytes(got.upstreamPayloads[0], "model").String())
+	require.Equal(t, "gpt-5.6-terra", gjson.GetBytes(got.upstreamPayloads[1], "model").String())
+
+	require.Len(t, got.logs, 2)
+	require.Equal(t, "sol", got.logs[0].Model)
+	require.NotNil(t, got.logs[0].UpstreamModel)
+	require.Equal(t, "gpt-5.6-sol", *got.logs[0].UpstreamModel)
+	require.NotNil(t, got.logs[0].ModelMappingChain)
+	require.Equal(t, "sol→gpt-5.6-sol", *got.logs[0].ModelMappingChain)
+	require.InDelta(t, 40e-6, got.logs[0].TotalCost, 1e-12,
+		"the in-flight turn must retain the channel-mapped billing model used when it was sent")
+
+	require.Equal(t, "sol", got.logs[1].Model)
+	require.NotNil(t, got.logs[1].UpstreamModel)
+	require.Equal(t, "gpt-5.6-terra", *got.logs[1].UpstreamModel)
+	require.NotNil(t, got.logs[1].ModelMappingChain)
+	require.Equal(t, "sol→gpt-5.6-terra", *got.logs[1].ModelMappingChain)
+	require.InDelta(t, got.logs[1].TotalCost*2.5, got.logs[0].TotalCost, 1e-12,
+		"the next turn must use the updated channel mapping")
+}
+
+func TestOpenAIResponsesWebSocket_CtxPoolAppliesPerTurnMappingAndPreservesRequestedModel(t *testing.T) {
+	got := runOpenAIResponsesWebSocketUsageLogCase(t, openAIResponsesWSUsageLogCase{
+		firstPayload:       `{"type":"response.create","model":"gpt-5.6-sol","stream":false}`,
+		secondPayload:      `{"type":"response.create","model":"gpt-5.6-terra","stream":false}`,
+		ingressMode:        service.OpenAIWSIngressModeCtxPool,
+		billingModelSource: service.BillingModelSourceRequested,
+		channelMapping: map[string]string{
+			"gpt-5.6-terra": "gpt-5.6-sol",
+		},
+	})
+
+	require.Len(t, got.upstreamPayloads, 2)
+	require.Equal(t, "gpt-5.6-sol", gjson.GetBytes(got.upstreamPayloads[0], "model").String())
+	require.Equal(t, "gpt-5.6-sol", gjson.GetBytes(got.upstreamPayloads[1], "model").String())
+	require.Len(t, got.clientEvents, 2)
+	require.Equal(t, "gpt-5.6-sol", gjson.GetBytes(got.clientEvents[0], "response.model").String())
+	require.Equal(t, "gpt-5.6-terra", gjson.GetBytes(got.clientEvents[1], "response.model").String())
+
+	require.Len(t, got.logs, 2)
+	require.Equal(t, "gpt-5.6-sol", got.logs[0].RequestedModel)
+	require.Nil(t, got.logs[0].ModelMappingChain)
+	require.InDelta(t, 40e-6, got.logs[0].TotalCost, 1e-12)
+	require.Equal(t, "gpt-5.6-terra", got.logs[1].RequestedModel)
+	require.NotNil(t, got.logs[1].ModelMappingChain)
+	require.Equal(t, "gpt-5.6-terra→gpt-5.6-sol", *got.logs[1].ModelMappingChain)
+	require.InDelta(t, 16e-6, got.logs[1].TotalCost, 1e-12,
+		"BillingModelSourceRequested must use the client model before channel mapping")
+}
+
+func TestOpenAIWSTurnBillingModelPreservesImagePricingModel(t *testing.T) {
+	tests := []struct {
+		name             string
+		resultModel      string
+		mapping          service.ChannelMappingResult
+		requestedModel   string
+		upstreamModel    string
+		wantBillingModel string
+	}{
+		{
+			name:             "upstream billing preserves image model",
+			resultModel:      "gpt-image-2",
+			mapping:          service.ChannelMappingResult{BillingModelSource: service.BillingModelSourceUpstream},
+			requestedModel:   "gpt-5.6-sol",
+			upstreamModel:    "gpt-5.6-sol",
+			wantBillingModel: "gpt-image-2",
+		},
+		{
+			name:             "unmapped channel preserves image model",
+			resultModel:      "gpt-image-2",
+			mapping:          service.ChannelMappingResult{MappedModel: "gpt-5.6-sol", BillingModelSource: service.BillingModelSourceChannelMapped},
+			requestedModel:   "gpt-5.6-sol",
+			upstreamModel:    "gpt-5.6-sol",
+			wantBillingModel: "gpt-image-2",
+		},
+		{
+			name:             "requested source overrides image model",
+			resultModel:      "gpt-image-2",
+			mapping:          service.ChannelMappingResult{BillingModelSource: service.BillingModelSourceRequested},
+			requestedModel:   "public-image-alias",
+			upstreamModel:    "gpt-5.6-sol",
+			wantBillingModel: "public-image-alias",
+		},
+		{
+			name:             "mapped channel source overrides image model",
+			resultModel:      "gpt-image-2",
+			mapping:          service.ChannelMappingResult{MappedModel: "priced-channel-model", BillingModelSource: service.BillingModelSourceChannelMapped},
+			requestedModel:   "public-image-alias",
+			upstreamModel:    "gpt-5.6-sol",
+			wantBillingModel: "priced-channel-model",
+		},
+		{
+			name:             "text turn falls back to upstream model",
+			mapping:          service.ChannelMappingResult{BillingModelSource: service.BillingModelSourceUpstream},
+			requestedModel:   "public-alias",
+			upstreamModel:    "gpt-5.6-sol",
+			wantBillingModel: "gpt-5.6-sol",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := &service.OpenAIForwardResult{BillingModel: tt.resultModel}
+			require.Equal(t, tt.wantBillingModel, openAIWSTurnBillingModel(result, tt.mapping, tt.requestedModel, tt.upstreamModel))
+		})
+	}
+}
+
+func TestShouldReportOpenAIWSProxyAccountFailure(t *testing.T) {
+	t.Run("unsupported client model switch does not penalize account", func(t *testing.T) {
+		err := fmt.Errorf("wrapped ingress turn: %w", newOpenAIWSUnsupportedModelSwitchError("gpt-unsupported"))
+		require.False(t, shouldReportOpenAIWSProxyAccountFailure(err))
+
+		var closeErr *service.OpenAIWSClientCloseError
+		require.ErrorAs(t, err, &closeErr)
+		require.Equal(t, coderws.StatusPolicyViolation, closeErr.StatusCode())
+		require.Equal(t, "model switch requires reconnect", closeErr.Reason())
+	})
+
+	t.Run("upstream policy violation still penalizes account", func(t *testing.T) {
+		err := service.NewOpenAIWSClientCloseError(
+			coderws.StatusPolicyViolation,
+			"upstream websocket authentication failed",
+			errors.New("upstream rejected credentials"),
+		)
+		require.True(t, shouldReportOpenAIWSProxyAccountFailure(err))
+	})
+
+	t.Run("generic proxy failure still penalizes account", func(t *testing.T) {
+		require.True(t, shouldReportOpenAIWSProxyAccountFailure(errors.New("upstream websocket read failed")))
+	})
+}
+
 func TestSetOpenAIClientTransportHTTP(t *testing.T) {
 	gin.SetMode(gin.TestMode)
 
@@ -1516,14 +1741,22 @@ func newOpenAIWSHandlerTestServer(t *testing.T, h *OpenAIGatewayHandler, subject
 }
 
 type openAIResponsesWSUsageLogCase struct {
-	firstPayload   string
-	userAgent      *string
-	channelMapping map[string]string
+	firstPayload              string
+	secondPayload             string
+	userAgent                 *string
+	ingressMode               string
+	channelMapping            map[string]string
+	billingModelSource        string
+	accountModelMapping       map[string]any
+	afterFirstUpstreamRequest func(channelSvc *service.ChannelService) error
 }
 
 type openAIResponsesWSUsageLogResult struct {
 	log                  *service.UsageLog
+	logs                 []*service.UsageLog
 	upstreamFirstPayload []byte
+	upstreamPayloads     [][]byte
+	clientEvents         [][]byte
 }
 
 type openAIWSUsageHandlerAccountRepoStub struct {
@@ -1579,6 +1812,40 @@ func (u *openAIHTTPPassthroughFailoverUpstream) calls() []int64 {
 	return append([]int64(nil), u.accountIDs...)
 }
 
+type openAIHTTPPassthroughSSERateLimitUpstream struct {
+	service.HTTPUpstream
+	mu         sync.Mutex
+	accountIDs []int64
+}
+
+func (u *openAIHTTPPassthroughSSERateLimitUpstream) Do(_ *http.Request, _ string, accountID int64, _ int) (*http.Response, error) {
+	u.mu.Lock()
+	u.accountIDs = append(u.accountIDs, accountID)
+	u.mu.Unlock()
+	body := strings.Join([]string{
+		"event: response.created",
+		`data: {"type":"response.created","response":{"id":"resp_rate_limited"}}`,
+		"",
+		"event: response.failed",
+		`data: {"type":"response.failed","response":{"id":"resp_rate_limited","status":"failed","error":{"code":"rate_limit_exceeded","message":"Concurrency limit exceeded for account, please retry later"}}}`,
+		"",
+	}, "\n")
+	return &http.Response{
+		StatusCode: http.StatusOK,
+		Header: http.Header{
+			"Content-Type": []string{"text/event-stream"},
+			"Retry-After":  []string{"1"},
+		},
+		Body: io.NopCloser(strings.NewReader(body)),
+	}, nil
+}
+
+func (u *openAIHTTPPassthroughSSERateLimitUpstream) calls() []int64 {
+	u.mu.Lock()
+	defer u.mu.Unlock()
+	return append([]int64(nil), u.accountIDs...)
+}
+
 func (s *openAIWSFailoverHandlerAccountRepoStub) ListSchedulableByPlatform(ctx context.Context, platform string) ([]service.Account, error) {
 	out := make([]service.Account, 0, len(s.accounts))
 	for _, account := range s.accounts {
@@ -1633,12 +1900,50 @@ func (s *openAIWSUsageHandlerUsageLogRepoStub) Create(ctx context.Context, log *
 
 type openAIWSUsageHandlerChannelRepoStub struct {
 	service.ChannelRepository
+	mu             sync.Mutex
 	channels       []service.Channel
 	groupPlatforms map[int64]string
 }
 
 func (s *openAIWSUsageHandlerChannelRepoStub) ListAll(ctx context.Context) ([]service.Channel, error) {
-	return s.channels, nil
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	out := make([]service.Channel, 0, len(s.channels))
+	for i := range s.channels {
+		out = append(out, *s.channels[i].Clone())
+	}
+	return out, nil
+}
+
+func (s *openAIWSUsageHandlerChannelRepoStub) GetByID(ctx context.Context, id int64) (*service.Channel, error) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	for i := range s.channels {
+		if s.channels[i].ID == id {
+			return s.channels[i].Clone(), nil
+		}
+	}
+	return nil, service.ErrChannelNotFound
+}
+
+func (s *openAIWSUsageHandlerChannelRepoStub) Update(ctx context.Context, channel *service.Channel) error {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	for i := range s.channels {
+		if s.channels[i].ID == channel.ID {
+			s.channels[i] = *channel.Clone()
+			return nil
+		}
+	}
+	return service.ErrChannelNotFound
+}
+
+func (s *openAIWSUsageHandlerChannelRepoStub) GetGroupIDs(ctx context.Context, channelID int64) ([]int64, error) {
+	channel, err := s.GetByID(ctx, channelID)
+	if err != nil {
+		return nil, err
+	}
+	return append([]int64(nil), channel.GroupIDs...), nil
 }
 
 func (s *openAIWSUsageHandlerChannelRepoStub) GetGroupPlatforms(ctx context.Context, groupIDs []int64) (map[int64]string, error) {
@@ -1741,6 +2046,88 @@ func TestOpenAIResponses_APIKeyPassthroughPool5xxRetriesThenExhaustsMaxSwitches(
 	require.Equal(t, "Upstream service temporarily unavailable", gjson.GetBytes(rec.Body.Bytes(), "error.message").String())
 }
 
+func TestOpenAIResponses_APIKeyPassthroughSSERateLimitUsesConfiguredPoolRetry(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	groupID := int64(4204)
+	accounts := []service.Account{
+		{
+			ID: 9912, Name: "pool-sse-rate-limit", Platform: service.PlatformOpenAI,
+			Type: service.AccountTypeAPIKey, Status: service.StatusActive, Schedulable: true, Priority: 1,
+			Credentials: map[string]any{
+				"api_key":                      "sk-pool",
+				"base_url":                     "https://api.example.test",
+				"pool_mode":                    true,
+				"pool_mode_retry_count":        float64(1),
+				"pool_mode_retry_status_codes": []any{float64(http.StatusTooManyRequests)},
+			},
+			Extra: map[string]any{"openai_passthrough": true},
+		},
+	}
+	cfg := &config.Config{RunMode: config.RunModeSimple}
+	cfg.Default.RateMultiplier = 1
+	cfg.Security.URLAllowlist.Enabled = false
+	cfg.Gateway.MaxAccountSwitches = 1
+
+	accountRepo := &openAIWSFailoverHandlerAccountRepoStub{accounts: accounts}
+	upstream := &openAIHTTPPassthroughSSERateLimitUpstream{}
+	billingCacheSvc := service.NewBillingCacheService(nil, nil, nil, nil, nil, nil, cfg, nil)
+	t.Cleanup(billingCacheSvc.Stop)
+	gatewaySvc := service.NewOpenAIGatewayService(
+		accountRepo,
+		nil,
+		nil,
+		nil,
+		nil,
+		nil,
+		nil,
+		cfg,
+		nil,
+		nil,
+		service.NewBillingService(cfg, nil),
+		nil,
+		billingCacheSvc,
+		upstream,
+		&service.DeferredService{},
+		nil,
+		nil,
+		nil,
+		nil,
+		nil,
+		nil,
+		nil,
+	)
+	h := NewOpenAIGatewayHandler(
+		gatewaySvc,
+		service.NewConcurrencyService(nil),
+		billingCacheSvc,
+		service.NewAPIKeyService(nil, nil, nil, nil, nil, nil, cfg),
+		nil,
+		nil,
+		nil,
+		nil,
+		cfg,
+	)
+
+	rec := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(rec)
+	c.Request = httptest.NewRequest(http.MethodPost, "/openai/v1/responses", strings.NewReader(`{"model":"gpt-5.6-sol","input":"hello","stream":true}`))
+	c.Request.Header.Set("Content-Type", "application/json")
+	c.Set(string(middleware.ContextKeyAPIKey), &service.APIKey{
+		ID: 1804, GroupID: &groupID,
+		User:  &service.User{ID: 1704, Status: service.StatusActive},
+		Group: &service.Group{ID: groupID, Platform: service.PlatformOpenAI, Status: service.StatusActive},
+	})
+	c.Set(string(middleware.ContextKeyUser), middleware.AuthSubject{UserID: 1704, Concurrency: 0})
+
+	h.Responses(c)
+
+	require.Equal(t, []int64{9912, 9912}, upstream.calls())
+	require.Equal(t, http.StatusTooManyRequests, rec.Code)
+	require.Equal(t, "1", rec.Header().Get("Retry-After"))
+	require.Equal(t, "rate_limit_error", gjson.GetBytes(rec.Body.Bytes(), "error.type").String())
+	require.Equal(t, "Upstream rate limit exceeded, please retry later", gjson.GetBytes(rec.Body.Bytes(), "error.message").String())
+}
+
 func TestOpenAIResponsesWebSocket_FailoverOnUpstreamUsageLimitEvent(t *testing.T) {
 	gin.SetMode(gin.TestMode)
 
@@ -2147,8 +2534,13 @@ func runOpenAIResponsesWebSocketUsageLogCase(t *testing.T, tc openAIResponsesWSU
 	t.Helper()
 	gin.SetMode(gin.TestMode)
 
-	upstreamPayloadCh := make(chan []byte, 1)
+	turnCount := 1
+	if strings.TrimSpace(tc.secondPayload) != "" {
+		turnCount = 2
+	}
+	upstreamPayloadCh := make(chan []byte, turnCount)
 	upstreamErrCh := make(chan error, 1)
+	var channelSvc *service.ChannelService
 	upstreamServer := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
 		conn, err := coderws.Accept(w, r, &coderws.AcceptOptions{
 			CompressionMode: coderws.CompressionContextTakeover,
@@ -2161,29 +2553,39 @@ func runOpenAIResponsesWebSocketUsageLogCase(t *testing.T, tc openAIResponsesWSU
 			_ = conn.CloseNow()
 		}()
 
-		readCtx, cancelRead := context.WithTimeout(r.Context(), 3*time.Second)
-		msgType, payload, readErr := conn.Read(readCtx)
-		cancelRead()
-		if readErr != nil {
-			upstreamErrCh <- readErr
-			return
-		}
-		if msgType != coderws.MessageText && msgType != coderws.MessageBinary {
-			upstreamErrCh <- errors.New("unexpected upstream websocket message type")
-			return
-		}
-		upstreamPayloadCh <- payload
+		for turn := 1; turn <= turnCount; turn++ {
+			readCtx, cancelRead := context.WithTimeout(r.Context(), 3*time.Second)
+			msgType, payload, readErr := conn.Read(readCtx)
+			cancelRead()
+			if readErr != nil {
+				upstreamErrCh <- readErr
+				return
+			}
+			if msgType != coderws.MessageText && msgType != coderws.MessageBinary {
+				upstreamErrCh <- errors.New("unexpected upstream websocket message type")
+				return
+			}
+			upstreamPayloadCh <- payload
+			if turn == 1 && tc.afterFirstUpstreamRequest != nil {
+				if callbackErr := tc.afterFirstUpstreamRequest(channelSvc); callbackErr != nil {
+					upstreamErrCh <- callbackErr
+					return
+				}
+			}
 
-		writeCtx, cancelWrite := context.WithTimeout(r.Context(), 3*time.Second)
-		writeErr := conn.Write(writeCtx, coderws.MessageText, []byte(
-			`{"type":"response.completed","response":{"id":"resp_usage_e2e","model":"gpt-5.4","usage":{"input_tokens":2,"output_tokens":1}}}`,
-		))
-		cancelWrite()
-		if writeErr != nil {
-			upstreamErrCh <- writeErr
-			return
+			response := fmt.Sprintf(
+				`{"type":"response.completed","response":{"id":"resp_usage_e2e_%d","model":%q,"usage":{"input_tokens":2,"output_tokens":1}}}`,
+				turn,
+				gjson.GetBytes(payload, "model").String(),
+			)
+			writeCtx, cancelWrite := context.WithTimeout(r.Context(), 3*time.Second)
+			writeErr := conn.Write(writeCtx, coderws.MessageText, []byte(response))
+			cancelWrite()
+			if writeErr != nil {
+				upstreamErrCh <- writeErr
+				return
+			}
 		}
-		_ = conn.Close(coderws.StatusNormalClosure, "done")
 		upstreamErrCh <- nil
 	}))
 	defer upstreamServer.Close()
@@ -2198,14 +2600,18 @@ func runOpenAIResponsesWebSocketUsageLogCase(t *testing.T, tc openAIResponsesWSU
 		Schedulable: true,
 		Concurrency: 1,
 		Credentials: map[string]any{
-			"api_key":  "sk-test",
-			"base_url": upstreamServer.URL,
+			"api_key":       "sk-test",
+			"base_url":      upstreamServer.URL,
+			"model_mapping": tc.accountModelMapping,
 		},
 		Extra: map[string]any{
 			"openai_apikey_responses_websockets_v2_enabled": true,
 			"openai_apikey_responses_websockets_v2_mode":    service.OpenAIWSIngressModePassthrough,
 		},
 	}
+	if strings.TrimSpace(tc.ingressMode) != "" {
+		account.Extra["openai_apikey_responses_websockets_v2_mode"] = tc.ingressMode
+	}
 
 	cfg := &config.Config{}
 	cfg.RunMode = config.RunModeSimple
@@ -2221,17 +2627,17 @@ func runOpenAIResponsesWebSocketUsageLogCase(t *testing.T, tc openAIResponsesWSU
 	cfg.Gateway.OpenAIWS.WriteTimeoutSeconds = 3
 
 	accountRepo := &openAIWSUsageHandlerAccountRepoStub{account: account}
-	usageRepo := &openAIWSUsageHandlerUsageLogRepoStub{created: make(chan *service.UsageLog, 1)}
+	usageRepo := &openAIWSUsageHandlerUsageLogRepoStub{created: make(chan *service.UsageLog, turnCount)}
 
-	var channelSvc *service.ChannelService
 	if len(tc.channelMapping) > 0 {
 		channelSvc = service.NewChannelService(&openAIWSUsageHandlerChannelRepoStub{
 			channels: []service.Channel{{
-				ID:           7701,
-				Name:         "openai-ws-e2e-channel",
-				Status:       service.StatusActive,
-				GroupIDs:     []int64{groupID},
-				ModelMapping: map[string]map[string]string{service.PlatformOpenAI: tc.channelMapping},
+				ID:                 7701,
+				Name:               "openai-ws-e2e-channel",
+				Status:             service.StatusActive,
+				GroupIDs:           []int64{groupID},
+				ModelMapping:       map[string]map[string]string{service.PlatformOpenAI: tc.channelMapping},
+				BillingModelSource: tc.billingModelSource,
 			}},
 			groupPlatforms: map[int64]string{groupID: service.PlatformOpenAI},
 		}, nil, nil, nil)
@@ -2314,26 +2720,44 @@ func runOpenAIResponsesWebSocketUsageLogCase(t *testing.T, tc openAIResponsesWSU
 	cancelWrite()
 	require.NoError(t, err)
 
-	readCtx, cancelRead := context.WithTimeout(context.Background(), 3*time.Second)
-	_, event, err := clientConn.Read(readCtx)
-	cancelRead()
-	require.NoError(t, err)
-	require.Equal(t, "response.completed", gjson.GetBytes(event, "type").String())
+	clientEvents := make([][]byte, 0, turnCount)
+	readCompleted := func() {
+		readCtx, cancelRead := context.WithTimeout(context.Background(), 3*time.Second)
+		_, event, readErr := clientConn.Read(readCtx)
+		cancelRead()
+		require.NoError(t, readErr)
+		require.Equal(t, "response.completed", gjson.GetBytes(event, "type").String())
+		clientEvents = append(clientEvents, append([]byte(nil), event...))
+	}
+	readCompleted()
+	if turnCount == 2 {
+		writeCtx, cancelWrite = context.WithTimeout(context.Background(), 3*time.Second)
+		err = clientConn.Write(writeCtx, coderws.MessageText, []byte(tc.secondPayload))
+		cancelWrite()
+		require.NoError(t, err)
+		readCompleted()
+	}
 	_ = clientConn.Close(coderws.StatusNormalClosure, "done")
 
-	var usageLog *service.UsageLog
-	select {
-	case usageLog = <-usageRepo.created:
-		require.NotNil(t, usageLog)
-	case <-time.After(3 * time.Second):
-		t.Fatal("等待 WebSocket usage log 写入超时")
+	usageLogs := make([]*service.UsageLog, 0, turnCount)
+	for len(usageLogs) < turnCount {
+		select {
+		case usageLog := <-usageRepo.created:
+			require.NotNil(t, usageLog)
+			usageLogs = append(usageLogs, usageLog)
+		case <-time.After(3 * time.Second):
+			t.Fatal("等待 WebSocket usage log 写入超时")
+		}
 	}
 
-	var upstreamFirstPayload []byte
-	select {
-	case upstreamFirstPayload = <-upstreamPayloadCh:
-	case <-time.After(3 * time.Second):
-		t.Fatal("等待上游 WebSocket 首帧超时")
+	upstreamPayloads := make([][]byte, 0, turnCount)
+	for len(upstreamPayloads) < turnCount {
+		select {
+		case payload := <-upstreamPayloadCh:
+			upstreamPayloads = append(upstreamPayloads, payload)
+		case <-time.After(3 * time.Second):
+			t.Fatal("等待上游 WebSocket 请求帧超时")
+		}
 	}
 
 	select {
@@ -2344,8 +2768,11 @@ func runOpenAIResponsesWebSocketUsageLogCase(t *testing.T, tc openAIResponsesWSU
 	}
 
 	return openAIResponsesWSUsageLogResult{
-		log:                  usageLog,
-		upstreamFirstPayload: upstreamFirstPayload,
+		log:                  usageLogs[0],
+		logs:                 usageLogs,
+		upstreamFirstPayload: upstreamPayloads[0],
+		upstreamPayloads:     upstreamPayloads,
+		clientEvents:         clientEvents,
 	}
 }
 

backend/internal/handler/openai_gateway_reasoning_failover.go

@@ -0,0 +1,62 @@
+package handler
+
+import (
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"go.uber.org/zap"
+)
+
+// openAIPassthroughFailoverState tracks whether this forwarding loop has attempted
+// an OpenAI passthrough account. Once it has, every subsequent non-passthrough
+// attempt must use a sanitized body because the immutable canonical body can carry
+// provider-specific encrypted reasoning produced by that passthrough upstream.
+type openAIPassthroughFailoverState struct {
+	passthroughSeen bool
+}
+
+// deriveOpenAIForwardAttemptBody returns the request body for the upcoming forward
+// attempt against account. It always derives from the immutable canonical body and
+// only removes encrypted reasoning input item(s) when the loop has already attempted
+// a passthrough account and the upcoming account is non-passthrough. This remains
+// sticky across retries and additional non-passthrough accounts. Attempts before
+// any passthrough account, and all passthrough attempts, forward the canonical body
+// unchanged. The canonical slice is never mutated.
+//
+// This method is invoked exactly once per forward attempt, immediately before the
+// Forward call, and advances the failover state as a side effect.
+func (h *OpenAIGatewayHandler) deriveOpenAIForwardAttemptBody(
+	reqLog *zap.Logger,
+	canonicalBody []byte,
+	account *service.Account,
+	state *openAIPassthroughFailoverState,
+) []byte {
+	currentPassthrough := account.IsOpenAIPassthroughEnabled()
+	if currentPassthrough {
+		state.passthroughSeen = true
+		return canonicalBody
+	}
+	if !state.passthroughSeen {
+		return canonicalBody
+	}
+
+	sanitized, changed, err := service.SanitizeOpenAICrossModeFailoverReasoning(canonicalBody)
+	if err != nil {
+		if reqLog != nil {
+			reqLog.Warn("openai.failover_cross_mode_reasoning_sanitize_failed",
+				zap.Int64("account_id", account.ID),
+				zap.Error(err),
+			)
+		}
+		return canonicalBody
+	}
+	if !changed {
+		return canonicalBody
+	}
+	if reqLog != nil {
+		reqLog.Info("openai.failover_cross_mode_reasoning_stripped",
+			zap.Int64("account_id", account.ID),
+			zap.Bool("account_passthrough", currentPassthrough),
+			zap.Bool("passthrough_seen", true),
+		)
+	}
+	return sanitized
+}

backend/internal/handler/openai_gateway_reasoning_failover_test.go

@@ -0,0 +1,146 @@
+package handler
+
+import (
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/stretchr/testify/require"
+	"github.com/tidwall/gjson"
+)
+
+// kiroReasoningCanonicalBody mirrors a native OpenAI Responses request whose input
+// carries a provider-specific encrypted reasoning item (Kiro shape: encrypted_content
+// coupled with an id and summary), plus surrounding message items.
+const kiroReasoningCanonicalBody = `{"model":"gpt-5.1","stream":false,"input":[` +
+	`{"type":"message","role":"user","content":"hello"},` +
+	`{"type":"reasoning","id":"rs_kiro_abc123","encrypted_content":"ENC_BLOB","summary":[{"type":"summary_text","text":"thinking"}]},` +
+	`{"type":"message","role":"assistant","content":"hi"}` +
+	`]}`
+
+func newOpenAIPassthroughAccount(id int64, passthrough bool) *service.Account {
+	extra := map[string]any{"openai_passthrough": passthrough}
+	return &service.Account{
+		ID:       id,
+		Platform: service.PlatformOpenAI,
+		Type:     service.AccountTypeAPIKey,
+		Extra:    extra,
+	}
+}
+
+func reasoningItemCount(t *testing.T, body []byte) int {
+	t.Helper()
+	count := 0
+	gjson.GetBytes(body, "input").ForEach(func(_, item gjson.Result) bool {
+		if item.Get("type").String() == "reasoning" {
+			count++
+		}
+		return true
+	})
+	return count
+}
+
+// TestDeriveOpenAIForwardAttemptBody_CrossModeStripsKiroReasoning drives the real
+// account-switch/request-body seam used inside OpenAIGatewayHandler.Responses: the
+// same deriveOpenAIForwardAttemptBody call the failover loop makes, fed real
+// *service.Account values in sequence. It proves the first passthrough attempt gets
+// the untouched reasoning item, and after a failover to a non-passthrough account
+// the second attempt body has the entire reasoning item removed — while the
+// immutable canonical body is never mutated.
+func TestDeriveOpenAIForwardAttemptBody_CrossModeStripsKiroReasoning(t *testing.T) {
+	h := &OpenAIGatewayHandler{}
+	canonical := []byte(kiroReasoningCanonicalBody)
+
+	kiro := newOpenAIPassthroughAccount(1, true)     // Kiro: openai_passthrough=true
+	bedrock := newOpenAIPassthroughAccount(2, false) // Bedrock Mantle: passthrough disabled
+
+	state := &openAIPassthroughFailoverState{}
+
+	// Attempt 1 — passthrough (Kiro): original reasoning item must survive intact.
+	firstBody := h.deriveOpenAIForwardAttemptBody(nil, canonical, kiro, state)
+	require.Equal(t, 1, reasoningItemCount(t, firstBody), "first passthrough attempt must keep the reasoning item")
+	require.Equal(t, "ENC_BLOB", gjson.GetBytes(firstBody, "input.1.encrypted_content").String())
+	require.Equal(t, "rs_kiro_abc123", gjson.GetBytes(firstBody, "input.1.id").String())
+	require.JSONEq(t, kiroReasoningCanonicalBody, string(firstBody), "first attempt body must equal the canonical body")
+
+	// Attempt 2 — failover switches to non-passthrough (Bedrock): the whole
+	// provider-specific encrypted reasoning item (id/summary/encrypted_content) is dropped.
+	secondBody := h.deriveOpenAIForwardAttemptBody(nil, canonical, bedrock, state)
+	require.Equal(t, 0, reasoningItemCount(t, secondBody), "cross-mode attempt must drop the reasoning item entirely")
+	require.False(t, gjson.GetBytes(secondBody, "input.#(encrypted_content)").Exists(), "no encrypted_content may remain")
+	require.NotContains(t, string(secondBody), "rs_kiro_abc123", "coupled reasoning id must be gone")
+	require.NotContains(t, string(secondBody), "ENC_BLOB")
+
+	// Surviving message items are preserved.
+	require.Equal(t, 2, int(gjson.GetBytes(secondBody, "input.#").Int()))
+	require.Equal(t, "hello", gjson.GetBytes(secondBody, "input.0.content").String())
+	require.Equal(t, "hi", gjson.GetBytes(secondBody, "input.1.content").String())
+
+	// The canonical body is immutable across attempts.
+	require.JSONEq(t, kiroReasoningCanonicalBody, string(canonical), "canonical forwardBody must never be mutated")
+}
+
+// TestDeriveOpenAIForwardAttemptBody_SameModePreservesReasoning proves that
+// non-passthrough attempts before any passthrough attempt, passthrough-family
+// failovers, and switching into passthrough forward the canonical reasoning item
+// unchanged.
+func TestDeriveOpenAIForwardAttemptBody_SameModePreservesReasoning(t *testing.T) {
+	h := &OpenAIGatewayHandler{}
+	canonical := []byte(kiroReasoningCanonicalBody)
+
+	t.Run("non_passthrough_to_non_passthrough", func(t *testing.T) {
+		state := &openAIPassthroughFailoverState{}
+		first := h.deriveOpenAIForwardAttemptBody(nil, canonical, newOpenAIPassthroughAccount(10, false), state)
+		second := h.deriveOpenAIForwardAttemptBody(nil, canonical, newOpenAIPassthroughAccount(11, false), state)
+		require.Equal(t, 1, reasoningItemCount(t, first))
+		require.Equal(t, 1, reasoningItemCount(t, second), "Bedrock-family failover must preserve reasoning")
+		require.JSONEq(t, kiroReasoningCanonicalBody, string(second))
+	})
+
+	t.Run("passthrough_to_passthrough", func(t *testing.T) {
+		state := &openAIPassthroughFailoverState{}
+		first := h.deriveOpenAIForwardAttemptBody(nil, canonical, newOpenAIPassthroughAccount(20, true), state)
+		second := h.deriveOpenAIForwardAttemptBody(nil, canonical, newOpenAIPassthroughAccount(21, true), state)
+		require.Equal(t, 1, reasoningItemCount(t, first))
+		require.Equal(t, 1, reasoningItemCount(t, second), "Kiro-family failover must preserve reasoning")
+	})
+
+	t.Run("non_passthrough_to_passthrough", func(t *testing.T) {
+		state := &openAIPassthroughFailoverState{}
+		_ = h.deriveOpenAIForwardAttemptBody(nil, canonical, newOpenAIPassthroughAccount(30, false), state)
+		second := h.deriveOpenAIForwardAttemptBody(nil, canonical, newOpenAIPassthroughAccount(31, true), state)
+		require.Equal(t, 1, reasoningItemCount(t, second), "switching into passthrough must preserve reasoning")
+	})
+
+	t.Run("same_account_pool_retry", func(t *testing.T) {
+		state := &openAIPassthroughFailoverState{}
+		kiro := newOpenAIPassthroughAccount(40, true)
+		_ = h.deriveOpenAIForwardAttemptBody(nil, canonical, kiro, state)
+		retry := h.deriveOpenAIForwardAttemptBody(nil, canonical, kiro, state)
+		require.Equal(t, 1, reasoningItemCount(t, retry), "same-account retry must preserve reasoning")
+	})
+}
+
+// TestDeriveOpenAIForwardAttemptBody_SanitizationSticksAcrossBedrockRetries proves
+// that once a passthrough account has been attempted, all later non-passthrough
+// retries and failovers remain sanitized instead of restoring the canonical Kiro
+// reasoning item after the first Bedrock failure.
+func TestDeriveOpenAIForwardAttemptBody_SanitizationSticksAcrossBedrockRetries(t *testing.T) {
+	h := &OpenAIGatewayHandler{}
+	canonical := []byte(kiroReasoningCanonicalBody)
+	state := &openAIPassthroughFailoverState{}
+
+	kiro := newOpenAIPassthroughAccount(50, true)
+	bedrockA := newOpenAIPassthroughAccount(51, false)
+	bedrockB := newOpenAIPassthroughAccount(52, false)
+
+	first := h.deriveOpenAIForwardAttemptBody(nil, canonical, kiro, state)
+	second := h.deriveOpenAIForwardAttemptBody(nil, canonical, bedrockA, state)
+	retry := h.deriveOpenAIForwardAttemptBody(nil, canonical, bedrockA, state)
+	nextAccount := h.deriveOpenAIForwardAttemptBody(nil, canonical, bedrockB, state)
+
+	require.Equal(t, 1, reasoningItemCount(t, first))
+	require.Equal(t, 0, reasoningItemCount(t, second))
+	require.Equal(t, 0, reasoningItemCount(t, retry), "same Bedrock account retry must remain sanitized")
+	require.Equal(t, 0, reasoningItemCount(t, nextAccount), "later Bedrock account must remain sanitized")
+	require.JSONEq(t, kiroReasoningCanonicalBody, string(canonical), "canonical forwardBody must never be mutated")
+}

backend/internal/handler/openai_images.go

@@ -91,6 +91,8 @@ func (h *OpenAIGatewayHandler) Images(c *gin.Context) {
 		zap.Bool("stream", parsed.Stream),
 		zap.Bool("multipart", parsed.Multipart),
 		zap.String("capability", string(parsed.RequiredCapability)),
+		zap.String("img_quality", parsed.Quality),
+		zap.String("img_size", parsed.Size),
 	)
 
 	if !service.GroupAllowsImageGeneration(apiKey.Group) {
@@ -146,6 +148,7 @@ func (h *OpenAIGatewayHandler) Images(c *gin.Context) {
 
 	maxAccountSwitches := h.maxAccountSwitches
 	switchCount := 0
+	profitVetoCount := 0
 	failedAccountIDs := make(map[int64]struct{})
 	sameAccountRetryCount := make(map[int64]int)
 	var lastFailoverErr *service.UpstreamFailoverError
@@ -219,8 +222,16 @@ func (h *OpenAIGatewayHandler) Images(c *gin.Context) {
 		reqLog.Debug("openai.images.account_selected", zap.Int64("account_id", account.ID), zap.String("account_name", account.Name))
 		setOpsSelectedAccount(c, account.ID, account.Platform)
 
-		accountReleaseFunc, acquired := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, parsed.Stream, &streamStarted, reqLog)
-		if !acquired {
+		accountReleaseFunc, slotResult := h.acquireResponsesAccountSlot(c, apiKey.GroupID, sessionHash, selection, parsed.Stream, &streamStarted, reqLog)
+		if slotResult == openAISlotAcquireProfitVetoed {
+			// Images 调度不装利润门，此分支实际不可达；防御性排除重选并受同一否决上限约束。
+			if !recordOpenAIProfitVeto(failedAccountIDs, account.ID, &profitVetoCount) {
+				h.handleOpenAIProfitVetoExhausted(c, streamStarted, reqLog, profitVetoCount)
+				return
+			}
+			continue
+		}
+		if slotResult != openAISlotAcquireOK {
 			return
 		}
 
@@ -374,6 +385,7 @@ func (h *OpenAIGatewayHandler) Images(c *gin.Context) {
 		if result != nil {
 			upstreamModel = result.UpstreamModel
 		}
+		sessionID := service.ExtractClientSessionID(c)
 		h.submitMandatoryUsageRecordTask(c.Request.Context(), func(ctx context.Context) {
 			if err := h.gatewayService.RecordUsage(ctx, &service.OpenAIRecordUsageInput{
 				Result:             result,
@@ -388,6 +400,7 @@ func (h *OpenAIGatewayHandler) Images(c *gin.Context) {
 				RequestPayloadHash: requestPayloadHash,
 				APIKeyService:      h.apiKeyService,
 				QuotaPlatform:      quotaPlatform,
+				SessionID:          sessionID,
 				ChannelUsageFields: clientRequestedUsageFields(c, channelMapping, requestModel, upstreamModel),
 			}); err != nil {
 				logger.L().With(

backend/internal/handler/openai_images_failover_test.go

@@ -12,11 +12,14 @@ import (
 	"testing"
 
 	"github.com/Wei-Shaw/sub2api/internal/config"
+	"github.com/Wei-Shaw/sub2api/internal/pkg/logger"
 	middleware2 "github.com/Wei-Shaw/sub2api/internal/server/middleware"
 	"github.com/Wei-Shaw/sub2api/internal/service"
 	"github.com/gin-gonic/gin"
 	"github.com/stretchr/testify/require"
 	"github.com/tidwall/gjson"
+	"go.uber.org/zap"
+	"go.uber.org/zap/zaptest/observer"
 )
 
 type openAIImagesFailoverAccountRepo struct {
@@ -154,8 +157,10 @@ func TestOpenAIGatewayHandlerImages_ServerErrorFailsOverAndReturnsClearErrorWhen
 	)
 	handler.maxAccountSwitches = 10
 
-	body := []byte(`{"model":"gpt-image-2","prompt":"draw a cat"}`)
-	req := httptest.NewRequest(http.MethodPost, "/v1/images/generations", bytes.NewReader(body))
+	body := []byte(`{"model":"gpt-image-2","prompt":"draw a cat","quality":"high","size":"1536x1024"}`)
+	core, observedLogs := observer.New(zap.DebugLevel)
+	requestCtx := logger.IntoContext(context.Background(), zap.New(core))
+	req := httptest.NewRequest(http.MethodPost, "/v1/images/generations", bytes.NewReader(body)).WithContext(requestCtx)
 	req.Header.Set("Content-Type", "application/json")
 	rec := httptest.NewRecorder()
 	c, _ := gin.CreateTestContext(rec)
@@ -173,6 +178,16 @@ func TestOpenAIGatewayHandlerImages_ServerErrorFailsOverAndReturnsClearErrorWhen
 
 	handler.Images(c)
 
+	accountSelectingLogs := observedLogs.FilterMessage("openai.images.account_selecting").All()
+	require.NotEmpty(t, accountSelectingLogs)
+	loggedFields := make(map[string]string)
+	for _, field := range accountSelectingLogs[0].Context {
+		loggedFields[field.Key] = field.String
+	}
+	require.Equal(t, "high", loggedFields["img_quality"])
+	require.Equal(t, "1536x1024", loggedFields["img_size"])
+	require.NotContains(t, loggedFields, "prompt")
+
 	require.Equal(t, []int64{1, 2}, upstream.calls())
 	require.Equal(t, http.StatusBadGateway, rec.Code)
 	require.Equal(t, "upstream_error", gjson.GetBytes(rec.Body.Bytes(), "error.type").String())

backend/internal/handler/openai_live.go

@@ -0,0 +1,236 @@
+package handler
+
+import (
+	"encoding/json"
+	"errors"
+	"io"
+	"net/http"
+	"net/url"
+	"strconv"
+	"strings"
+
+	"github.com/Wei-Shaw/sub2api/internal/pkg/ip"
+	middleware2 "github.com/Wei-Shaw/sub2api/internal/server/middleware"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	coderws "github.com/coder/websocket"
+	"github.com/gin-gonic/gin"
+	"github.com/tidwall/gjson"
+	"go.uber.org/zap"
+)
+
+func (h *OpenAIGatewayHandler) Live(c *gin.Context) {
+	apiKey, ok := middleware2.GetAPIKeyFromContext(c)
+	if !ok {
+		h.errorResponse(c, http.StatusUnauthorized, "authentication_error", "Invalid API key")
+		return
+	}
+	subject, ok := middleware2.GetAuthSubjectFromContext(c)
+	if !ok {
+		h.errorResponse(c, http.StatusInternalServerError, "api_error", "User context not found")
+		return
+	}
+	if apiKey.Group == nil || apiKey.Group.Platform != service.PlatformOpenAI {
+		h.errorResponse(c, http.StatusNotFound, "not_found_error", "Live is not supported for this platform")
+		return
+	}
+	if !liveEnabledForAPIKey(apiKey) {
+		h.errorResponse(c, http.StatusForbidden, "permission_error", "Live is not enabled for this group")
+		return
+	}
+	request, err := parseLiveCallRequest(c)
+	if err != nil {
+		h.errorResponse(c, http.StatusBadRequest, "invalid_request_error", err.Error())
+		return
+	}
+	model := strings.TrimSpace(gjson.GetBytes(request.Session, "model").String())
+	reqLog := requestLogger(
+		c,
+		"handler.openai_gateway.live",
+		zap.Int64("user_id", subject.UserID),
+		zap.Int64("api_key_id", apiKey.ID),
+		zap.Any("group_id", apiKey.GroupID),
+	)
+	if decision := h.checkSecurityAudit(
+		c,
+		reqLog,
+		apiKey,
+		subject,
+		service.ContentModerationProtocolOpenAIResponses,
+		model,
+		request.Session,
+	); decision != nil && !decision.AllowNextStage {
+		h.openAISecurityAuditError(c, decision)
+		return
+	}
+
+	subscription, _ := middleware2.GetSubscriptionFromContext(c)
+	if h.billingCacheService == nil {
+		h.errorResponse(c, http.StatusServiceUnavailable, "api_error", "Billing service unavailable")
+		return
+	}
+	if err := h.billingCacheService.CheckBillingEligibility(
+		c.Request.Context(),
+		apiKey.User,
+		apiKey,
+		apiKey.Group,
+		subscription,
+		service.QuotaPlatform(c.Request.Context(), apiKey),
+	); err != nil {
+		status, code, message, retryAfter := billingErrorDetails(err)
+		if retryAfter > 0 {
+			c.Header("Retry-After", strconv.Itoa(retryAfter))
+		}
+		h.errorResponse(c, status, code, message)
+		return
+	}
+
+	userRelease, acquired, err := h.concurrencyHelper.TryAcquireUserSlot(
+		c.Request.Context(),
+		subject.UserID,
+		subject.Concurrency,
+	)
+	if err != nil {
+		h.errorResponse(c, http.StatusServiceUnavailable, "api_error", "Live concurrency unavailable")
+		return
+	}
+	if !acquired {
+		h.errorResponse(c, http.StatusTooManyRequests, "rate_limit_error", "Live concurrency limit reached")
+		return
+	}
+	defer userRelease()
+
+	identity := liveCallIdentity(c, apiKey, subject.UserID, subscription)
+	created, err := h.gatewayService.CreateLiveCall(c.Request.Context(), request, identity, subject.Concurrency)
+	if err != nil {
+		h.writeLiveCreateError(c, err)
+		return
+	}
+	c.Header("Location", liveSidebandLocation(c.FullPath(), created.CallID))
+	c.Data(http.StatusOK, "application/sdp", created.SDP)
+}
+
+func parseLiveCallRequest(c *gin.Context) (*service.LiveCallRequest, error) {
+	contentType := strings.ToLower(c.GetHeader("Content-Type"))
+	if strings.HasPrefix(contentType, "multipart/form-data") {
+		sdp := c.PostForm("sdp")
+		session := json.RawMessage(c.PostForm("session"))
+		request := &service.LiveCallRequest{SDP: sdp, Session: session}
+		if err := service.ValidateLiveCallRequest(request); err != nil {
+			return nil, err
+		}
+		return request, nil
+	}
+	var request service.LiveCallRequest
+	decoder := json.NewDecoder(c.Request.Body)
+	if err := decoder.Decode(&request); err != nil {
+		return nil, errors.New("request body must be valid JSON")
+	}
+	if err := decoder.Decode(&struct{}{}); !errors.Is(err, io.EOF) {
+		return nil, errors.New("request body must contain one JSON object")
+	}
+	if err := service.ValidateLiveCallRequest(&request); err != nil {
+		return nil, err
+	}
+	return &request, nil
+}
+
+func liveSidebandLocation(fullPath, callID string) string {
+	prefix := "/v1/live/"
+	if strings.HasPrefix(fullPath, "/backend-api/codex/") {
+		prefix = "/backend-api/codex/"
+	}
+	return prefix + url.PathEscape(callID)
+}
+
+func liveCallIdentity(
+	c *gin.Context,
+	apiKey *service.APIKey,
+	userID int64,
+	subscription *service.UserSubscription,
+) service.LiveCallIdentity {
+	var subscriptionID *int64
+	if subscription != nil {
+		value := subscription.ID
+		subscriptionID = &value
+	}
+	return service.LiveCallIdentity{
+		APIKeyID:        apiKey.ID,
+		UserID:          userID,
+		GroupID:         apiKey.GroupID,
+		SubscriptionID:  subscriptionID,
+		UserAgent:       c.GetHeader("User-Agent"),
+		IPAddress:       ip.GetClientIP(c),
+		InboundEndpoint: GetInboundEndpoint(c),
+	}
+}
+
+func (h *OpenAIGatewayHandler) writeLiveCreateError(c *gin.Context, err error) {
+	switch {
+	case errors.Is(err, service.ErrLiveConcurrencyFull):
+		h.errorResponse(c, http.StatusTooManyRequests, "rate_limit_error", "Live concurrency limit reached")
+	case errors.Is(err, service.ErrLiveUnavailable):
+		h.errorResponse(c, http.StatusServiceUnavailable, "api_error", "Live is unavailable")
+	default:
+		var attestationErr *service.LiveAttestationUnavailableError
+		if errors.As(err, &attestationErr) {
+			h.errorResponse(c, http.StatusServiceUnavailable, "api_error", attestationErr.Error())
+			return
+		}
+		var upstreamErr *service.UpstreamFailoverError
+		if errors.As(err, &upstreamErr) && upstreamErr.StatusCode >= 400 && upstreamErr.StatusCode < 500 {
+			h.errorResponse(c, upstreamErr.StatusCode, "invalid_request_error", "Live upstream rejected the request")
+			return
+		}
+		h.errorResponse(c, http.StatusBadGateway, "api_error", "Live upstream request failed")
+	}
+}
+
+func (h *OpenAIGatewayHandler) LiveSideband(c *gin.Context) {
+	apiKey, ok := middleware2.GetAPIKeyFromContext(c)
+	if !ok {
+		h.errorResponse(c, http.StatusUnauthorized, "authentication_error", "Invalid API key")
+		return
+	}
+	subject, ok := middleware2.GetAuthSubjectFromContext(c)
+	if !ok {
+		h.errorResponse(c, http.StatusInternalServerError, "api_error", "User context not found")
+		return
+	}
+	if !liveEnabledForAPIKey(apiKey) {
+		h.errorResponse(c, http.StatusForbidden, "permission_error", "Live is not enabled for this group")
+		return
+	}
+	identity := service.LiveCallIdentity{
+		APIKeyID: apiKey.ID,
+		UserID:   subject.UserID,
+		GroupID:  apiKey.GroupID,
+	}
+	record, err := h.gatewayService.GetLiveCallForIdentity(c.Request.Context(), c.Param("call_id"), identity)
+	if err != nil {
+		if errors.Is(err, service.ErrLiveIdentityMismatch) {
+			h.errorResponse(c, http.StatusForbidden, "permission_error", "Live call belongs to another identity")
+			return
+		}
+		h.errorResponse(c, http.StatusNotFound, "not_found_error", "Live call not found")
+		return
+	}
+	downstream, err := coderws.Accept(c.Writer, c.Request, &coderws.AcceptOptions{
+		InsecureSkipVerify: true,
+	})
+	if err != nil {
+		return
+	}
+	defer func() { _ = downstream.CloseNow() }()
+	if err := h.gatewayService.ProxyLiveSideband(c.Request.Context(), record, downstream); err != nil {
+		_ = downstream.Close(coderws.StatusInternalError, "live sideband closed")
+		return
+	}
+	_ = downstream.Close(coderws.StatusNormalClosure, "")
+}
+
+func liveEnabledForAPIKey(apiKey *service.APIKey) bool {
+	return apiKey != nil &&
+		apiKey.Group != nil &&
+		apiKey.Group.Platform == service.PlatformOpenAI &&
+		apiKey.Group.AllowLive
+}

backend/internal/handler/openai_live_test.go

@@ -0,0 +1,118 @@
+package handler
+
+import (
+	"bytes"
+	"encoding/json"
+	"mime/multipart"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/require"
+)
+
+func TestParseLiveCallRequestMultipartPreservesSession(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	session := `{"model":"gpt-live-test","delegation":{"type":"client"},"instructions":"你好"}`
+	var body bytes.Buffer
+	writer := multipart.NewWriter(&body)
+	require.NoError(t, writer.WriteField("sdp", "v=0\r\n"))
+	require.NoError(t, writer.WriteField("session", session))
+	require.NoError(t, writer.Close())
+
+	request := httptest.NewRequest("POST", "/v1/live", &body)
+	request.Header.Set("Content-Type", writer.FormDataContentType())
+	context, _ := gin.CreateTestContext(httptest.NewRecorder())
+	context.Request = request
+
+	parsed, err := parseLiveCallRequest(context)
+	require.NoError(t, err)
+	require.Equal(t, "v=0\r\n", parsed.SDP)
+	require.JSONEq(t, session, string(parsed.Session))
+	require.Equal(t, "client", jsonPathString(t, parsed.Session, "delegation", "type"))
+}
+
+func TestParseLiveCallRequestJSONPreservesSessionWithoutDelegation(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	body := `{"sdp":"v=0\\r\\n","session":{"model":"gpt-live-test","instructions":"standalone"}}`
+	request := httptest.NewRequest("POST", "/backend-api/codex/realtime/calls", bytes.NewBufferString(body))
+	request.Header.Set("Content-Type", "application/json")
+	context, _ := gin.CreateTestContext(httptest.NewRecorder())
+	context.Request = request
+
+	parsed, err := parseLiveCallRequest(context)
+	require.NoError(t, err)
+	require.NotContains(t, string(parsed.Session), "delegation")
+	require.Equal(t, "standalone", jsonPathString(t, parsed.Session, "instructions"))
+}
+
+func TestParseLiveCallRequestRejectsInvalidJSONShape(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	testCases := []string{
+		`{"session":{"type":"quicksilver"}}`,
+		`{"sdp":"v=0\\r\\n","session":[]}`,
+		`{"sdp":"v=0\\r\\n","session":null}`,
+		`{"sdp":"v=0\\r\\n","session":{"type":"quicksilver"}} {}`,
+	}
+	for _, body := range testCases {
+		request := httptest.NewRequest("POST", "/backend-api/codex/realtime/calls", bytes.NewBufferString(body))
+		request.Header.Set("Content-Type", "application/json")
+		context, _ := gin.CreateTestContext(httptest.NewRecorder())
+		context.Request = request
+		_, err := parseLiveCallRequest(context)
+		require.Error(t, err)
+	}
+}
+
+func TestLiveSidebandLocationMatchesCreateRoute(t *testing.T) {
+	require.Equal(t, "/v1/live/call_123", liveSidebandLocation("/v1/live", "call_123"))
+	require.Equal(
+		t,
+		"/backend-api/codex/call_123",
+		liveSidebandLocation("/backend-api/codex/realtime/calls", "call_123"),
+	)
+}
+
+func TestLiveEnabledForAPIKey(t *testing.T) {
+	require.False(t, liveEnabledForAPIKey(nil))
+	require.False(t, liveEnabledForAPIKey(&service.APIKey{}))
+	require.False(t, liveEnabledForAPIKey(&service.APIKey{
+		Group: &service.Group{Platform: service.PlatformOpenAI},
+	}))
+	require.False(t, liveEnabledForAPIKey(&service.APIKey{
+		Group: &service.Group{Platform: service.PlatformAnthropic, AllowLive: true},
+	}))
+	require.True(t, liveEnabledForAPIKey(&service.APIKey{
+		Group: &service.Group{Platform: service.PlatformOpenAI, AllowLive: true},
+	}))
+}
+
+func TestLiveAttestationErrorIsExplicit(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	recorder := httptest.NewRecorder()
+	context, _ := gin.CreateTestContext(recorder)
+
+	(&OpenAIGatewayHandler{}).writeLiveCreateError(context, &service.LiveAttestationUnavailableError{
+		Reason: "Live attestation is only supported when Sub2API runs on macOS",
+	})
+
+	require.Equal(t, http.StatusServiceUnavailable, recorder.Code)
+	require.Contains(t, recorder.Body.String(), "Sub2API runs on macOS")
+}
+
+func jsonPathString(t *testing.T, raw json.RawMessage, keys ...string) string {
+	t.Helper()
+	var value any
+	require.NoError(t, json.Unmarshal(raw, &value))
+	current := value
+	for _, key := range keys {
+		object, ok := current.(map[string]any)
+		require.True(t, ok)
+		current = object[key]
+	}
+	result, ok := current.(string)
+	require.True(t, ok)
+	return result
+}

backend/internal/handler/openai_profit_slot_recheck_test.go

@@ -0,0 +1,150 @@
+//go:build unit
+
+package handler
+
+// 槽位终检与生图跳门回归（handler 半程）：
+//   - 槽位获取成功后的利润终检：越线账号释放槽位并要求调用方排除重选，
+//     不写响应、不绑定粘连；
+//   - openAIResponsesRequiredCapability 的生图意图映射钉死（scheduler 的
+//     跳门条件依赖 CapabilityResponses ⇔ 显式生图意图这一耦合）。
+
+import (
+	"context"
+	"net/http/httptest"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/Wei-Shaw/sub2api/internal/pkg/ctxkey"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/require"
+	"go.uber.org/zap"
+)
+
+type profitCountingConcurrencyCache struct {
+	fakeConcurrencyCache
+	accountReleases atomic.Int64
+}
+
+func (c *profitCountingConcurrencyCache) ReleaseAccountSlot(context.Context, int64, string) error {
+	c.accountReleases.Add(1)
+	return nil
+}
+
+func profitSlotTestAccount(id int64, rate float64) *service.Account {
+	now := time.Now()
+	return &service.Account{
+		ID:             id,
+		Platform:       service.PlatformOpenAI,
+		Type:           service.AccountTypeAPIKey,
+		Status:         service.StatusActive,
+		Schedulable:    true,
+		Concurrency:    2,
+		RateMultiplier: &rate,
+		Extra: map[string]any{
+			"upstream_billing_probe": map[string]any{
+				"status":      service.UpstreamBillingProbeStatusOK,
+				"received_at": now.Add(-time.Minute),
+				"fresh_until": now.Add(30 * time.Minute),
+				"data": map[string]any{
+					"billing_scope":            "token",
+					"resolved_rate_multiplier": rate,
+					"peak_rate_enabled":        false,
+				},
+			},
+		},
+	}
+}
+
+func profitSlotTestContext(t *testing.T, gw *service.OpenAIGatewayService, groupID int64, suppress bool) context.Context {
+	t.Helper()
+	group := &service.Group{
+		ID:                   groupID,
+		Platform:             service.PlatformOpenAI,
+		Status:               service.StatusActive,
+		Hydrated:             true,
+		RateMultiplier:       1.0,
+		SubscriptionType:     service.SubscriptionTypeStandard,
+		ProfitControlEnabled: true,
+		ProfitMinMargin:      0.5,
+	}
+	base := context.WithValue(context.Background(), ctxkey.Group, group)
+	if suppress {
+		base = service.WithOpenAIProfitControlSuppressed(base)
+	}
+	ctx, pricingAt := gw.WithOpenAIRequestPricingContext(base, &groupID)
+	require.False(t, pricingAt.IsZero())
+	return ctx
+}
+
+func TestAcquireResponsesAccountSlotProfitRecheck(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	gw := &service.OpenAIGatewayService{}
+	groupID := int64(50)
+
+	newHandler := func(cache *profitCountingConcurrencyCache) *OpenAIGatewayHandler {
+		return &OpenAIGatewayHandler{
+			gatewayService:    gw,
+			concurrencyHelper: NewConcurrencyHelper(service.NewConcurrencyService(cache), SSEPingFormatClaude, 0),
+		}
+	}
+	newSelection := func(account *service.Account) *service.AccountSelectionResult {
+		return &service.AccountSelectionResult{
+			Account:  account,
+			Acquired: false,
+			WaitPlan: &service.AccountWaitPlan{AccountID: account.ID, MaxConcurrency: 2, Timeout: time.Second, MaxWaiting: 2},
+		}
+	}
+
+	t.Run("veto releases slot and requests reschedule without writing response", func(t *testing.T) {
+		cache := &profitCountingConcurrencyCache{}
+		h := newHandler(cache)
+		w := httptest.NewRecorder()
+		c, _ := gin.CreateTestContext(w)
+		c.Request = httptest.NewRequest("POST", "/v1/responses", nil).WithContext(profitSlotTestContext(t, gw, groupID, false))
+		streamStarted := false
+
+		release, result := h.acquireResponsesAccountSlot(c, &groupID, "", newSelection(profitSlotTestAccount(1, 0.8)), false, &streamStarted, zap.NewNop())
+		require.Equal(t, openAISlotAcquireProfitVetoed, result)
+		require.Nil(t, release)
+		require.Zero(t, w.Body.Len(), "利润终检否决不得写出任何响应")
+		require.Equal(t, int64(1), cache.accountReleases.Load(), "否决后必须立即释放已获取的槽位")
+	})
+
+	t.Run("qualifying account acquires normally", func(t *testing.T) {
+		cache := &profitCountingConcurrencyCache{}
+		h := newHandler(cache)
+		w := httptest.NewRecorder()
+		c, _ := gin.CreateTestContext(w)
+		c.Request = httptest.NewRequest("POST", "/v1/responses", nil).WithContext(profitSlotTestContext(t, gw, groupID, false))
+		streamStarted := false
+
+		release, result := h.acquireResponsesAccountSlot(c, &groupID, "", newSelection(profitSlotTestAccount(2, 0.3)), false, &streamStarted, zap.NewNop())
+		require.Equal(t, openAISlotAcquireOK, result)
+		require.NotNil(t, release)
+		release()
+	})
+
+	t.Run("image intent suppression keeps official behavior", func(t *testing.T) {
+		cache := &profitCountingConcurrencyCache{}
+		h := newHandler(cache)
+		w := httptest.NewRecorder()
+		c, _ := gin.CreateTestContext(w)
+		c.Request = httptest.NewRequest("POST", "/v1/responses", nil).WithContext(profitSlotTestContext(t, gw, groupID, true))
+		streamStarted := false
+
+		release, result := h.acquireResponsesAccountSlot(c, &groupID, "", newSelection(profitSlotTestAccount(3, 0.8)), false, &streamStarted, zap.NewNop())
+		require.Equal(t, openAISlotAcquireOK, result, "生图意图跳门：过贵账号照常获取（图片边界不装门）")
+		require.NotNil(t, release)
+		release()
+	})
+}
+
+// scheduler 跳门条件依赖"CapabilityResponses 仅在显式生图意图时被要求"这一
+// 映射；后续若扩展该 capability 的用途，本测试失败提示同步收窄跳门条件。
+func TestOpenAIResponsesRequiredCapabilityPinsImageIntentMapping(t *testing.T) {
+	require.Equal(t, service.OpenAIEndpointCapabilityResponses, openAIResponsesRequiredCapability(true, service.PlatformOpenAI))
+	require.Equal(t, service.OpenAIEndpointCapabilityChatCompletions, openAIResponsesRequiredCapability(false, service.PlatformOpenAI))
+	require.Equal(t, service.OpenAIEndpointCapabilityChatCompletions, openAIResponsesRequiredCapability(true, service.PlatformGrok))
+}

backend/internal/handler/openai_profit_veto_budget_test.go

@@ -0,0 +1,74 @@
+package handler
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+// TestRecordOpenAIProfitVetoBounded 钉死 OpenAI 侧选号循环的利润否决预算：
+// 账号被加入排除集，且第 maxProfitVetoAttempts 次否决返回 false 要求终止。
+//
+// 上限是必需的：acquireResponsesAccountSlot 的 WaitPlan 分支先阻塞排队
+// （sticky 45s / fallback 30s）拿到槽位才做利润终检，无上限重选会把单次请求
+// 的延迟放大到 N × WaitPlan.Timeout。
+func TestRecordOpenAIProfitVetoBounded(t *testing.T) {
+	failed := make(map[int64]struct{})
+	count := 0
+
+	for i := int64(1); i < int64(maxProfitVetoAttempts); i++ {
+		require.True(t, recordOpenAIProfitVeto(failed, i, &count), "第 %d 次否决应允许继续重选", i)
+		require.Contains(t, failed, i, "否决的账号必须进入本请求排除集")
+	}
+
+	require.False(t,
+		recordOpenAIProfitVeto(failed, int64(maxProfitVetoAttempts), &count),
+		"第 %d 次否决应耗尽预算并要求终止", maxProfitVetoAttempts)
+	require.Equal(t, maxProfitVetoAttempts, count)
+	require.Len(t, failed, maxProfitVetoAttempts)
+}
+
+// TestOpenAIProfitVetoLoopTerminates 模拟「选号 → 抢槽 → 利润终检否决 → 重选」
+// 循环：整池越线时必须在常数步内终止，而不是把候选池逐个排队一遍。
+func TestOpenAIProfitVetoLoopTerminates(t *testing.T) {
+	failed := make(map[int64]struct{})
+	count := 0
+	terminated := false
+	iterations := 0
+
+	// 候选池远大于否决上限：无上限时会对每个账号各排队一次。
+	for accountID := int64(1); accountID <= 500; accountID++ {
+		iterations++
+		if _, excluded := failed[accountID]; excluded {
+			continue
+		}
+		if !recordOpenAIProfitVeto(failed, accountID, &count) {
+			terminated = true
+			break
+		}
+	}
+
+	require.True(t, terminated, "整池越线时循环必须由否决预算终止")
+	require.Equal(t, maxProfitVetoAttempts, iterations)
+}
+
+// TestProfitVetoBudgetSharedWithFailoverState 钉死两条路径（FailoverState 与
+// OpenAI 独立计数器）使用同一上限语义，避免日后单边漂移。
+func TestProfitVetoBudgetSharedWithFailoverState(t *testing.T) {
+	fs := NewFailoverState(10, false)
+	failed := make(map[int64]struct{})
+	count := 0
+
+	var fsStoppedAt, openAIStoppedAt int
+	for i := int64(1); i <= int64(maxProfitVetoAttempts)+5; i++ {
+		if fsStoppedAt == 0 && fs.RecordProfitVeto(i) == FailoverExhausted {
+			fsStoppedAt = fs.ProfitVetoCount()
+		}
+		if openAIStoppedAt == 0 && !recordOpenAIProfitVeto(failed, i, &count) {
+			openAIStoppedAt = count
+		}
+	}
+
+	require.Equal(t, maxProfitVetoAttempts, fsStoppedAt)
+	require.Equal(t, fsStoppedAt, openAIStoppedAt, "两条路径的利润否决上限必须一致")
+}

backend/internal/handler/openai_ws_turn_pricing_test.go

@@ -0,0 +1,34 @@
+package handler
+
+import (
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+)
+
+// TestOpenAIWSTurnPricingZeroValue 钉死 WS turn 定价的零值语义：
+// 没有 turn 起始回调的 ingress 模式（ws_v2 passthrough 只实现 AfterTurn）
+// 必须让 pricingAt 保持零，由 RecordUsage 回退到记录时刻。
+//
+// 反例（本 PR 引入的回归）：用建连时刻初始化，会把透传连接的所有 turn 钉死在
+// 建连时的高峰因子——客户端峰前一分钟建连并保活，整条连接就按谷价结算。
+func TestOpenAIWSTurnPricingZeroValue(t *testing.T) {
+	var p openAIWSTurnPricing
+	require.True(t, p.current().IsZero(),
+		"未经 turn 起始回调冻结时必须保持零值，交由 RecordUsage 回退记录时刻")
+}
+
+// TestOpenAIWSTurnPricingFreezePerTurn 钉死每个 turn 的 BeforeTurn 都会覆盖
+// 上一个 turn 的定价时刻：长连接跨峰谷时后续 turn 不得沿用旧时刻。
+func TestOpenAIWSTurnPricingFreezePerTurn(t *testing.T) {
+	var p openAIWSTurnPricing
+	turn1 := time.Now().Add(-time.Hour)
+	turn2 := time.Now()
+
+	p.freeze(turn1)
+	require.Equal(t, turn1, p.current())
+
+	p.freeze(turn2)
+	require.Equal(t, turn2, p.current(), "后续 turn 必须使用自己的定价时刻")
+}

backend/internal/handler/passkey_handler.go

@@ -0,0 +1,254 @@
+package handler
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"io"
+	"net/http"
+	"strconv"
+	"strings"
+
+	infraerrors "github.com/Wei-Shaw/sub2api/internal/pkg/errors"
+	"github.com/Wei-Shaw/sub2api/internal/pkg/response"
+	middleware2 "github.com/Wei-Shaw/sub2api/internal/server/middleware"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/gin-gonic/gin"
+)
+
+type PasskeyHandler struct {
+	passkeys    *service.PasskeyService
+	authService *service.AuthService
+	settingSvc  *service.SettingService
+}
+
+func NewPasskeyHandler(
+	passkeys *service.PasskeyService,
+	authService *service.AuthService,
+	settingService *service.SettingService,
+) *PasskeyHandler {
+	return &PasskeyHandler{
+		passkeys:    passkeys,
+		authService: authService,
+		settingSvc:  settingService,
+	}
+}
+
+type passkeyOptionsResponse struct {
+	SessionToken string `json:"session_token"`
+	Options      any    `json:"options"`
+}
+
+type passkeyFinishRequest struct {
+	SessionToken string          `json:"session_token" binding:"required"`
+	Name         string          `json:"name,omitempty"`
+	Credential   json.RawMessage `json:"credential" binding:"required"`
+}
+
+type passkeyRenameRequest struct {
+	Name string `json:"name" binding:"required"`
+}
+
+// passkeyPasswordRequest carries the account password that gates passkey
+// enrollment and revocation. Binding errors are tolerated: a missing or
+// malformed body yields an empty password, which the service rejects with a
+// precise PASSWORD_REQUIRED error.
+type passkeyPasswordRequest struct {
+	Password string `json:"password"`
+}
+
+func bindPasskeyPassword(c *gin.Context) string {
+	var req passkeyPasswordRequest
+	_ = c.ShouldBindJSON(&req)
+	return req.Password
+}
+
+const passkeyFinishBodyMaxBytes = 64 * 1024
+
+// BeginLogin starts a usernameless, discoverable-credential login ceremony.
+func (h *PasskeyHandler) BeginLogin(c *gin.Context) {
+	if !h.requirePasskeysEnabled(c) {
+		return
+	}
+	assertion, token, err := h.passkeys.BeginLogin(c.Request.Context())
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, passkeyOptionsResponse{SessionToken: token, Options: assertion})
+}
+
+// FinishLogin validates a passkey assertion and creates a normal Sub2API token
+// session. User verification is mandatory, so a successful passkey assertion
+// already supplies phishing-resistant multi-factor authentication and does not
+// enter the separate TOTP challenge flow.
+func (h *PasskeyHandler) FinishLogin(c *gin.Context) {
+	if !h.requirePasskeysEnabled(c) {
+		return
+	}
+	req, ok := bindPasskeyFinishRequest(c)
+	if !ok {
+		return
+	}
+	credentialRequest := cloneRequestWithJSON(c.Request, req.Credential)
+	user, err := h.passkeys.FinishLogin(c.Request.Context(), req.SessionToken, credentialRequest)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	if err = h.ensureBackendModeAllowsUser(c.Request.Context(), user); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	middleware2.SetAuditActor(c, user.ID, user.Email)
+	c.Set("auth_method", service.AuditAuthMethodPasskey)
+	h.authService.RecordSuccessfulLogin(c.Request.Context(), user.ID)
+	respondWithTokenPair(c, h.authService, user)
+}
+
+func (h *PasskeyHandler) BeginRegistration(c *gin.Context) {
+	if !h.requirePasskeysEnabled(c) {
+		return
+	}
+	subject, ok := middleware2.GetAuthSubjectFromContext(c)
+	if !ok {
+		response.Unauthorized(c, "User not authenticated")
+		return
+	}
+	creation, token, err := h.passkeys.BeginRegistration(c.Request.Context(), subject.UserID, bindPasskeyPassword(c))
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, passkeyOptionsResponse{SessionToken: token, Options: creation})
+}
+
+func (h *PasskeyHandler) FinishRegistration(c *gin.Context) {
+	if !h.requirePasskeysEnabled(c) {
+		return
+	}
+	subject, ok := middleware2.GetAuthSubjectFromContext(c)
+	if !ok {
+		response.Unauthorized(c, "User not authenticated")
+		return
+	}
+	req, valid := bindPasskeyFinishRequest(c)
+	if !valid {
+		return
+	}
+	credentialRequest := cloneRequestWithJSON(c.Request, req.Credential)
+	credential, err := h.passkeys.FinishRegistration(
+		c.Request.Context(),
+		subject.UserID,
+		req.SessionToken,
+		req.Name,
+		credentialRequest,
+	)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, credential)
+}
+
+func (h *PasskeyHandler) List(c *gin.Context) {
+	subject, ok := middleware2.GetAuthSubjectFromContext(c)
+	if !ok {
+		response.Unauthorized(c, "User not authenticated")
+		return
+	}
+	credentials, err := h.passkeys.List(c.Request.Context(), subject.UserID)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, credentials)
+}
+
+func (h *PasskeyHandler) Rename(c *gin.Context) {
+	subject, credentialID, ok := passkeyMutationTarget(c)
+	if !ok {
+		return
+	}
+	var req passkeyRenameRequest
+	if err := c.ShouldBindJSON(&req); err != nil || strings.TrimSpace(req.Name) == "" {
+		response.BadRequest(c, "Passkey name is required")
+		return
+	}
+	if err := h.passkeys.Rename(c.Request.Context(), subject.UserID, credentialID, req.Name); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, gin.H{"success": true})
+}
+
+func (h *PasskeyHandler) Delete(c *gin.Context) {
+	subject, credentialID, ok := passkeyMutationTarget(c)
+	if !ok {
+		return
+	}
+	if err := h.passkeys.Delete(c.Request.Context(), subject.UserID, credentialID, bindPasskeyPassword(c)); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, gin.H{"success": true})
+}
+
+func (h *PasskeyHandler) requirePasskeysEnabled(c *gin.Context) bool {
+	if h.settingSvc == nil {
+		response.ErrorFrom(c, service.ErrPasskeysDisabled)
+		return false
+	}
+	enabled, err := h.settingSvc.PasskeyEnabled(c.Request.Context())
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return false
+	}
+	if !enabled {
+		response.ErrorFrom(c, service.ErrPasskeysDisabled)
+	}
+	return enabled
+}
+
+func (h *PasskeyHandler) ensureBackendModeAllowsUser(ctx context.Context, user *service.User) error {
+	if err := ensureLoginUserActive(user); err != nil {
+		return err
+	}
+	if h.settingSvc == nil || !h.settingSvc.IsBackendModeEnabled(ctx) || user.IsAdmin() {
+		return nil
+	}
+	return infraerrors.Forbidden("BACKEND_MODE_ADMIN_ONLY", "Backend mode is active. Only admin login is allowed.")
+}
+
+func bindPasskeyFinishRequest(c *gin.Context) (*passkeyFinishRequest, bool) {
+	c.Request.Body = http.MaxBytesReader(c.Writer, c.Request.Body, passkeyFinishBodyMaxBytes)
+	var req passkeyFinishRequest
+	if err := c.ShouldBindJSON(&req); err != nil || len(req.Credential) == 0 {
+		response.BadRequest(c, "Invalid passkey response")
+		return nil, false
+	}
+	return &req, true
+}
+
+func cloneRequestWithJSON(original *http.Request, payload []byte) *http.Request {
+	request := original.Clone(original.Context())
+	request.Body = io.NopCloser(bytes.NewReader(payload))
+	request.ContentLength = int64(len(payload))
+	request.Header = original.Header.Clone()
+	request.Header.Set("Content-Type", "application/json")
+	return request
+}
+
+func passkeyMutationTarget(c *gin.Context) (middleware2.AuthSubject, int64, bool) {
+	subject, ok := middleware2.GetAuthSubjectFromContext(c)
+	if !ok {
+		response.Unauthorized(c, "User not authenticated")
+		return middleware2.AuthSubject{}, 0, false
+	}
+	credentialID, err := strconv.ParseInt(c.Param("id"), 10, 64)
+	if err != nil || credentialID <= 0 {
+		response.BadRequest(c, "Invalid passkey ID")
+		return middleware2.AuthSubject{}, 0, false
+	}
+	return subject, credentialID, true
+}

backend/internal/handler/passkey_handler_test.go

@@ -0,0 +1,101 @@
+package handler
+
+import (
+	"context"
+	"errors"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/config"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/require"
+)
+
+type passkeySwitchSettingRepo struct {
+	value string
+	err   error
+}
+
+func (r *passkeySwitchSettingRepo) Get(context.Context, string) (*service.Setting, error) {
+	return nil, service.ErrSettingNotFound
+}
+func (r *passkeySwitchSettingRepo) GetValue(context.Context, string) (string, error) {
+	return r.value, r.err
+}
+func (r *passkeySwitchSettingRepo) Set(context.Context, string, string) error { return nil }
+func (r *passkeySwitchSettingRepo) GetMultiple(context.Context, []string) (map[string]string, error) {
+	return map[string]string{}, nil
+}
+func (r *passkeySwitchSettingRepo) SetMultiple(context.Context, map[string]string) error {
+	return nil
+}
+func (r *passkeySwitchSettingRepo) GetAll(context.Context) (map[string]string, error) {
+	return map[string]string{}, nil
+}
+func (r *passkeySwitchSettingRepo) Delete(context.Context, string) error { return nil }
+
+func TestBindPasskeyFinishRequestRejectsOversizedBody(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	recorder := httptest.NewRecorder()
+	context, _ := gin.CreateTestContext(recorder)
+	context.Request = httptest.NewRequest(
+		http.MethodPost,
+		"/api/v1/auth/passkey/login/finish",
+		strings.NewReader(`{"credential":"`+strings.Repeat("x", passkeyFinishBodyMaxBytes)+`"}`),
+	)
+	context.Request.Header.Set("Content-Type", "application/json")
+
+	_, ok := bindPasskeyFinishRequest(context)
+	require.False(t, ok)
+	require.Equal(t, http.StatusBadRequest, recorder.Code)
+}
+
+func TestPasskeyBeginLoginRejectsDisabledAdminSwitch(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	repo := &passkeySwitchSettingRepo{value: "false"}
+	settings := service.NewSettingService(repo, &config.Config{
+		WebAuthn: config.WebAuthnConfig{Enabled: true},
+	})
+	handler := NewPasskeyHandler(nil, nil, settings)
+	recorder := httptest.NewRecorder()
+	ginContext, _ := gin.CreateTestContext(recorder)
+	ginContext.Request = httptest.NewRequest(http.MethodPost, "/api/v1/auth/passkey/login/begin", nil)
+
+	handler.BeginLogin(ginContext)
+
+	require.Equal(t, http.StatusForbidden, recorder.Code)
+	require.Contains(t, recorder.Body.String(), "PASSKEY_DISABLED")
+}
+
+func TestPasskeyBeginLoginReportsSettingStoreFailure(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	settings := service.NewSettingService(
+		&passkeySwitchSettingRepo{err: errors.New("database unavailable")},
+		&config.Config{WebAuthn: config.WebAuthnConfig{Enabled: true}},
+	)
+	handler := NewPasskeyHandler(nil, nil, settings)
+	recorder := httptest.NewRecorder()
+	ginContext, _ := gin.CreateTestContext(recorder)
+	ginContext.Request = httptest.NewRequest(http.MethodPost, "/api/v1/auth/passkey/login/begin", nil)
+
+	handler.BeginLogin(ginContext)
+
+	require.Equal(t, http.StatusInternalServerError, recorder.Code)
+	require.NotContains(t, recorder.Body.String(), "PASSKEY_DISABLED")
+}
+
+func TestPasskeyCredentialListRemainsAvailableWhenSignInDisabled(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	handler := NewPasskeyHandler(nil, nil, nil)
+	recorder := httptest.NewRecorder()
+	ginContext, _ := gin.CreateTestContext(recorder)
+	ginContext.Request = httptest.NewRequest(http.MethodGet, "/api/v1/user/passkeys", nil)
+
+	handler.List(ginContext)
+
+	require.Equal(t, http.StatusUnauthorized, recorder.Code)
+	require.NotContains(t, recorder.Body.String(), "PASSKEY_DISABLED")
+}

backend/internal/handler/setting_handler.go

@@ -52,6 +52,7 @@ func (h *SettingHandler) GetPublicSettings(c *gin.Context) {
 		PasswordResetEnabled:             settings.PasswordResetEnabled,
 		InvitationCodeEnabled:            settings.InvitationCodeEnabled,
 		TotpEnabled:                      settings.TotpEnabled,
+		PasskeyEnabled:                   settings.PasskeyEnabled,
 		LoginAgreementEnabled:            settings.LoginAgreementEnabled,
 		LoginAgreementMode:               settings.LoginAgreementMode,
 		LoginAgreementUpdatedAt:          settings.LoginAgreementUpdatedAt,
@@ -66,6 +67,7 @@ func (h *SettingHandler) GetPublicSettings(c *gin.Context) {
 		ContactInfo:                      settings.ContactInfo,
 		DocURL:                           settings.DocURL,
 		HomeContent:                      settings.HomeContent,
+		CompactHomeEnabled:               settings.CompactHomeEnabled,
 		HideCcsImportButton:              settings.HideCcsImportButton,
 		PurchaseSubscriptionEnabled:      settings.PurchaseSubscriptionEnabled,
 		PurchaseSubscriptionURL:          settings.PurchaseSubscriptionURL,
@@ -98,6 +100,9 @@ func (h *SettingHandler) GetPublicSettings(c *gin.Context) {
 
 		AvailableChannelsEnabled: settings.AvailableChannelsEnabled,
 
+		ModelPlazaEnabled:     settings.ModelPlazaEnabled,
+		ModelPlazaRequireAuth: settings.ModelPlazaRequireAuth,
+
 		AffiliateEnabled: settings.AffiliateEnabled,
 
 		RiskControlEnabled: settings.RiskControlEnabled,

backend/internal/handler/usage_record_task_fallback_test.go

@@ -0,0 +1,77 @@
+package handler
+
+import (
+	"context"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/Wei-Shaw/sub2api/internal/config"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/stretchr/testify/require"
+)
+
+// 本文件覆盖：worker 池已停止（进程关停窗口）时，计费任务不得静默丢失，
+// 必须降级为内联同步执行；显式配置的 drop/sample 溢出丢弃仍按配置语义保留。
+
+func newStoppedUsageRecordPoolForTest() *service.UsageRecordWorkerPool {
+	pool := service.NewUsageRecordWorkerPoolWithOptions(service.UsageRecordWorkerPoolOptions{
+		WorkerCount:    1,
+		QueueSize:      1,
+		TaskTimeout:    time.Second,
+		OverflowPolicy: config.UsageRecordOverflowPolicySync,
+	})
+	pool.Stop()
+	return pool
+}
+
+func TestGatewayHandlerSubmitUsageRecordTask_StoppedPoolFallsBackToSync(t *testing.T) {
+	h := &GatewayHandler{usageRecordWorkerPool: newStoppedUsageRecordPoolForTest()}
+
+	executed := false
+	h.submitUsageRecordTask(context.Background(), func(ctx context.Context) {
+		executed = true
+	})
+	require.True(t, executed, "池已停止时计费任务必须内联同步执行")
+}
+
+func TestOpenAIGatewayHandlerSubmitUsageRecordTask_StoppedPoolFallsBackToSync(t *testing.T) {
+	h := &OpenAIGatewayHandler{usageRecordWorkerPool: newStoppedUsageRecordPoolForTest()}
+
+	executed := false
+	h.submitUsageRecordTask(context.Background(), func(ctx context.Context) {
+		executed = true
+	})
+	require.True(t, executed, "池已停止时计费任务必须内联同步执行")
+}
+
+func TestGatewayHandlerSubmitUsageRecordTask_DropPolicyOverflowStillDrops(t *testing.T) {
+	pool := service.NewUsageRecordWorkerPoolWithOptions(service.UsageRecordWorkerPoolOptions{
+		WorkerCount:    1,
+		QueueSize:      1,
+		TaskTimeout:    time.Minute,
+		OverflowPolicy: config.UsageRecordOverflowPolicyDrop,
+	})
+	t.Cleanup(pool.Stop)
+	h := &GatewayHandler{usageRecordWorkerPool: pool}
+
+	started := make(chan struct{})
+	block := make(chan struct{})
+	t.Cleanup(func() { close(block) })
+	// 占满 worker 槽位后再填满队列，保证第三个任务触发溢出。
+	require.Equal(t, service.UsageRecordSubmitModeEnqueued, pool.Submit(func(ctx context.Context) {
+		close(started)
+		<-block
+	}))
+	<-started
+	require.Equal(t, service.UsageRecordSubmitModeEnqueued, pool.Submit(func(ctx context.Context) {
+		<-block
+	}))
+
+	var executed atomic.Bool
+	h.submitUsageRecordTask(context.Background(), func(ctx context.Context) {
+		executed.Store(true)
+	})
+	time.Sleep(50 * time.Millisecond)
+	require.False(t, executed.Load(), "drop 溢出策略是运维显式配置的取舍，不应被同步兜底覆盖")
+}

backend/internal/handler/user_handler_test.go

@@ -26,6 +26,9 @@ type userHandlerRepoStub struct {
 }
 
 func (s *userHandlerRepoStub) Create(context.Context, *service.User) error { return nil }
+func (s *userHandlerRepoStub) CreateWithEmailAliasGuard(context.Context, *service.User) error {
+	return nil
+}
 func (s *userHandlerRepoStub) GetByID(context.Context, int64) (*service.User, error) {
 	cloned := *s.user
 	return &cloned, nil
@@ -38,7 +41,7 @@ func (s *userHandlerRepoStub) GetFirstAdmin(context.Context) (*service.User, err
 	cloned := *s.user
 	return &cloned, nil
 }
-func (s *userHandlerRepoStub) Update(_ context.Context, user *service.User) error {
+func (s *userHandlerRepoStub) Update(_ context.Context, user *service.User, _ service.UserUpdateFields) error {
 	cloned := *user
 	s.user = &cloned
 	return nil
@@ -90,6 +93,14 @@ func (s *userHandlerRepoStub) UpdateConcurrency(context.Context, int64, int) err
 func (s *userHandlerRepoStub) BatchSetConcurrency(context.Context, []int64, int) (int, error) {
 	return 0, nil
 }
+
+func (s *userHandlerRepoStub) AdjustBalance(ctx context.Context, id int64, delta float64) (service.BalanceChange, error) {
+	panic("unexpected AdjustBalance call")
+}
+
+func (s *userHandlerRepoStub) SetBalance(ctx context.Context, id int64, value float64) (service.BalanceChange, error) {
+	panic("unexpected SetBalance call")
+}
 func (s *userHandlerRepoStub) BatchAddConcurrency(context.Context, []int64, int) (int, error) {
 	return 0, nil
 }
@@ -97,6 +108,9 @@ func (s *userHandlerRepoStub) BatchUpdateLimits(context.Context, []int64, *int,
 	return 0, nil
 }
 func (s *userHandlerRepoStub) ExistsByEmail(context.Context, string) (bool, error) { return false, nil }
+func (s *userHandlerRepoStub) ExistsByEmailAlias(context.Context, string) (bool, error) {
+	return false, nil
+}
 func (s *userHandlerRepoStub) RemoveGroupFromAllowedGroups(context.Context, int64) (int64, error) {
 	return 0, nil
 }
@@ -650,7 +664,11 @@ func TestUserHandlerUnbindIdentityRevokesAllUserSessionsWhenAuthServiceConfigure
 
 	require.Equal(t, http.StatusOK, recorder.Code)
 	require.Equal(t, []int64{23}, refreshTokenCache.revokedUserIDs)
-	require.Equal(t, int64(5), repo.user.TokenVersion)
+	// 撤销依赖的是 refresh session 清理，而不是 token_version：users 表没有这一列
+	// （见 resolvedTokenVersion，实际值由 email+password_hash 指纹推导），
+	// 所以此前"自增 TokenVersion 再整行写回"不持久化任何东西，
+	// 却会用旧快照覆盖并发写入的列。这里断言用户行未被改写。
+	require.Equal(t, int64(4), repo.user.TokenVersion)
 }
 
 func TestUserHandlerUnbindIdentityDoesNotRevokeSessionsWhenNothingWasUnbound(t *testing.T) {

backend/internal/handler/wire.go

@@ -179,9 +179,11 @@ func ProvideHandlers(
 	openaiGatewayHandler *OpenAIGatewayHandler,
 	settingHandler *SettingHandler,
 	totpHandler *TotpHandler,
+	passkeyHandler *PasskeyHandler,
 	paymentHandler *PaymentHandler,
 	paymentWebhookHandler *PaymentWebhookHandler,
 	availableChannelHandler *AvailableChannelHandler,
+	modelPlazaHandler *ModelPlazaHandler,
 	asyncImageHandler *AsyncImageHandler,
 	batchImageHandler *BatchImageHandler,
 	_ *service.IdempotencyCoordinator,
@@ -201,9 +203,11 @@ func ProvideHandlers(
 		OpenAIGateway:    openaiGatewayHandler,
 		Setting:          settingHandler,
 		Totp:             totpHandler,
+		Passkey:          passkeyHandler,
 		Payment:          paymentHandler,
 		PaymentWebhook:   paymentWebhookHandler,
 		AvailableChannel: availableChannelHandler,
+		ModelPlaza:       modelPlazaHandler,
 		AsyncImage:       asyncImageHandler,
 		BatchImage:       batchImageHandler,
 	}
@@ -223,10 +227,12 @@ var ProviderSet = wire.NewSet(
 	ProvideGatewayHandler,
 	ProvideOpenAIGatewayHandler,
 	NewTotpHandler,
+	NewPasskeyHandler,
 	ProvideSettingHandler,
 	NewPaymentHandler,
 	NewPaymentWebhookHandler,
 	NewAvailableChannelHandler,
+	NewModelPlazaHandler,
 	NewAsyncImageHandler,
 	ProvideBatchImageHandler,