Merge pull request #4428 from alfadb/fix/anthropic-haiku-full-mimicry

fix(gateway): 修复 Haiku OAuth 请求被归入额外用量

.github/workflows/backend-ci.yml

@@ -8,6 +8,15 @@ permissions:
   contents: read
 
 jobs:
+  shell:
+    runs-on: macos-15
+    steps:
+      - uses: actions/checkout@v6
+      - name: Check deployment scripts
+        run: |
+          /bin/bash -n deploy/apple-container.sh
+          /bin/bash deploy/tests/apple-container-test.sh
+
   test:
     runs-on: ubuntu-latest
     steps:

.gitignore

@@ -116,6 +116,8 @@ backend/.installed
 # 其他
 # ===================
 tests
+!deploy/tests/
+!deploy/tests/**
 CLAUDE.md
 .claude
 scripts
@@ -133,6 +135,7 @@ docs/*
 !docs/PAYMENT.md
 !docs/PAYMENT_CN.md
 !docs/ADMIN_PAYMENT_INTEGRATION_API.md
+!docs/ASYNC_IMAGE_TASKS.md
 !docs/legal/
 !docs/legal/*.md
 .serena/

README.md

@@ -163,6 +163,12 @@ Model authenticity: no content intervention or secondary filtering — experienc
 </td>
 </tr>
 
+<tr>
+<td width="180"><a href="http://aimzoon.com"><img src="assets/partners/logos/aimzoon.jpg" alt="aimzoon" width="150"></a></td>
+<td>Thanks to Aimzoon for sponsoring this project! <a href="http://aimzoon.com">Aimzoon</a> provides stable, cost-effective AI API access services, enabling developers to quickly connect popular AI services to coding tools such as Codex, Claude Code, and Gemini CLI. No complex configuration — faster onboarding, more stable calls, and lower costs. Ongoing promotions including discounted Codex rates and special pricing, with free trial credits upon registration, bringing AI coding into your daily workflow. <a href="http://aimzoon.com">Click here</a> to register and try it out!
+</td>
+</tr>
+
 </table>
 
 ## Overview
@@ -329,6 +335,7 @@ cd sub2api/deploy
 
 # 2. Copy environment configuration
 cp .env.example .env
+chmod 600 .env
 
 # 3. Edit configuration (generate secure passwords)
 nano .env
@@ -448,7 +455,23 @@ rm -rf data/ postgres_data/ redis_data/
 
 ---
 
-### Method 3: Build from Source
+### Method 3: Apple container (macOS)
+
+Apple-silicon Macs running macOS 26 can run the full Sub2API, PostgreSQL, and Redis stack with Apple `container` 1.1.0 or newer:
+
+```bash
+git clone https://github.com/Wei-Shaw/sub2api.git
+cd sub2api/deploy
+./apple-container.sh init
+./apple-container.sh up
+./apple-container.sh status
+```
+
+This is an operator-managed local workflow; Docker Compose remains the recommended production path. See [deploy/APPLE_CONTAINER.md](deploy/APPLE_CONTAINER.md) for lifecycle commands, persistence, upgrades, and runtime limitations.
+
+---
+
+### Method 4: Build from Source
 
 Build and run from source code for development or customization.
 
@@ -579,6 +602,34 @@ If you disable URL validation or response header filtering, harden your network
 - Enforce TLS-only outbound traffic
 - Strip sensitive upstream response headers at the proxy
 
+#### OpenAI Responses WebSocket ingress limits
+
+`gateway.openai_ws` bounds the lifetime and aggregate count of client-facing
+Responses WebSocket sessions. These safeguards apply independently from
+per-turn user and account concurrency slots, which are released between turns.
+
+```yaml
+gateway:
+  openai_ws:
+    # Total time to receive and decompress the first client message.
+    client_first_message_timeout_seconds: 30
+    # Close a client socket idle between completed turns; 0 disables this safeguard.
+    ingress_inter_turn_idle_timeout_seconds: 300
+    # Distributed API-key limit for live client ingress sessions; 0 disables it.
+    max_ingress_connections_per_api_key: 64
+```
+
+The first-message timeout is a total read deadline. Deployments that accept
+large contexts or image-heavy requests over slower links can raise it to
+120-300 seconds. It expires before HTTP bridge routing, so bridge mode does not
+override this limit.
+
+The connection cap is coordinated through Redis using a 60-second lease that
+is refreshed every 20 seconds. A process that cannot confirm a lease for a
+full lease lifetime closes its local WebSocket rather than continuing outside
+the global cap. Use `http_bridge` for client-WebSocket/upstream-HTTP operation
+when rolling out or mitigating upstream WebSocket issues.
+
 #### ⚠️ Important: Creating the Admin Account
 
 The initial admin account is **only created via the setup wizard** (served at `http://<host>:8080` on first run). The `default.admin_email` / `default.admin_password` fields in `config.yaml` are **not used** to create it — they exist in the template for historical reasons.
@@ -637,20 +688,26 @@ Simple Mode is designed for individual developers or internal teams who want qui
 
 ---
 
-## Grok / xAI OAuth Support
+## Asynchronous Image Tasks
 
-Sub2API supports Grok subscription accounts through xAI OAuth and forwards OpenAI-compatible Responses traffic to xAI.
+Long-running OpenAI/Grok image generation and editing can be submitted through `/v1/images/generations/async` or `/v1/images/edits/async`, then polled at `/v1/images/tasks/{task_id}` without holding a CDN connection open. See [Asynchronous Image Tasks](docs/ASYNC_IMAGE_TASKS.md) for request and response examples.
+
+---
+
+## Grok / xAI Support
+
+Sub2API supports both Grok subscription accounts through xAI OAuth and standard xAI API-key accounts. Both account types forward OpenAI-compatible Responses traffic to xAI.
 
 ### Supported Scope
 
 - Platform name: `grok`
-- Account type: OAuth subscription accounts
-- Public Responses targets: `/v1/responses`, `/responses`, and `/backend-api/codex/responses`, forwarded to `${XAI_BASE_URL:-https://api.x.ai/v1}/responses`
+- Account types: OAuth subscription accounts and xAI API-key accounts
+- Public Responses targets: `/v1/responses`, `/responses`, and `/backend-api/codex/responses`, forwarded to the Grok subscription proxy for OAuth accounts or `https://api.x.ai/v1/responses` for API-key accounts
 - Public Claude-compatible target: `/v1/messages`, converted to xAI Responses and returned as Anthropic Messages output for Claude CLI style clients
-- Public Chat Completions targets: `/v1/chat/completions` and `/chat/completions`, forwarded to `${XAI_BASE_URL:-https://api.x.ai/v1}/chat/completions`
+- Public Chat Completions targets: `/v1/chat/completions` and `/chat/completions`, forwarded to the account-type-specific xAI upstream
 - Codex CLI style Responses WebSocket ingress is accepted on the Responses targets and bridged to xAI HTTP/SSE Responses upstream
-- Initial text models: `grok-4.3`, `grok-build-0.1`, `grok-4.20-0309-reasoning`, `grok-4.20-0309-non-reasoning`, and `grok-4.20-multi-agent-0309`
-- Media targets for Grok groups: `/v1/images/generations`, `/images/generations`, `/v1/images/edits`, `/images/edits`, `/v1/videos/generations`, `/videos/generations`, `/v1/videos/{request_id}`, and `/videos/{request_id}`. Generation requests require the group image-generation permission.
+- Text models: `grok-4.5`, `grok-4.3`, `grok-build-0.1`, `grok-composer-2.5-fast`, `grok-4.20-0309-reasoning`, `grok-4.20-0309-non-reasoning`, and `grok-4.20-multi-agent-0309`
+- Media targets for Grok groups: `/v1/images/generations`, `/images/generations`, `/v1/images/edits`, `/images/edits`, `/v1/videos/generations`, `/videos/generations`, `/v1/videos/edits`, `/videos/edits`, `/v1/videos/extensions`, `/videos/extensions`, `/v1/videos/{request_id}`, and `/videos/{request_id}`. Generation, editing, and extension requests require the group image-generation permission.
 - Media models: `grok-imagine`, `grok-imagine-image-quality`, `grok-imagine-image`, `grok-imagine-edit`, `grok-imagine-video`, and `grok-imagine-video-1.5`
 - Out of scope for this provider: TTS, transcription, browser automation, cookies, and Grok web scraping
 
@@ -665,9 +722,10 @@ The Grok OAuth flow uses PKCE and does not require committing private secrets. T
 | `XAI_OAUTH_REDIRECT_URI` | `http://127.0.0.1:56121/callback` |
 | `XAI_OAUTH_AUTHORIZE_URL` | `https://auth.x.ai/oauth2/authorize` |
 | `XAI_OAUTH_TOKEN_URL` | `https://auth.x.ai/oauth2/token` |
-| `XAI_BASE_URL` | `https://api.x.ai/v1` |
+| `XAI_BASE_URL` | `https://api.x.ai/v1`; runtime-diagnostics override (account `base_url` controls request forwarding) |
+| `XAI_GROK_CLI_VERSION` | `0.2.93`; optional override for the client identity sent to `cli-chat-proxy.grok.com` |
 
-Administrators can create or reauthorize Grok accounts from the dashboard, or use the admin API:
+Administrators can create Grok OAuth or API-key accounts from the dashboard. OAuth authorization and reauthorization are also available through the admin API:
 
 | Endpoint | Purpose |
 |----------|---------|
@@ -676,13 +734,46 @@ Administrators can create or reauthorize Grok accounts from the dashboard, or us
 | `POST /api/v1/admin/grok/oauth/refresh-token` | Validate or refresh a Grok refresh token |
 | `POST /api/v1/admin/grok/accounts/:id/refresh` | Refresh an existing Grok account |
 
-Credential storage reuses the existing account JSON fields: `access_token`, `refresh_token`, `token_type`, `expires_at`, optional `email`, optional `subscription_tier`, and `entitlement_status`.
+OAuth credential storage reuses the existing account JSON fields: `access_token`, `refresh_token`, `token_type`, `expires_at`, `base_url`, optional `email`, optional `subscription_tier`, and `entitlement_status`. OAuth inference defaults to `https://cli-chat-proxy.grok.com/v1`; existing OAuth accounts that stored the old `https://api.x.ai/v1` default are redirected to the subscription proxy at runtime. Explicit custom upstreams remain unchanged.
+
+For API-key accounts, select **Grok → API Key** in the create-account dialog. The official base URL defaults to `https://api.x.ai/v1`; credentials use the existing `base_url` and `api_key` account fields. OAuth accounts continue to use the subscription flow above.
+
+### Grok Build CLI Configuration
+
+1. In the Sub2API admin dashboard, add either a `grok` OAuth account and complete xAI authorization, or add a Grok API-key account.
+2. Create a Grok group, attach the account to it, then create a Sub2API API key assigned to that group.
+3. In the user API-key page, click **Use Key** and select **Grok CLI**. The modal generates the correct file and base URL for macOS/Linux or Windows. It also provides an OpenCode configuration on the **OpenCode** tab.
+4. If configuring manually, save the following as `~/.grok/config.toml` (Windows: `%USERPROFILE%\.grok\config.toml`):
+
+```toml
+[models]
+default = "grok"
+web_search = "grok"
+
+[model."grok"]
+model = "grok-4.5"
+base_url = "https://your-sub2api.example.com/v1"
+name = "Grok 4.5"
+api_key = "sk-your-sub2api-key"
+api_backend = "responses"
+context_window = 1000000
+supports_backend_search = true
+```
+
+Back up an existing `config.toml` before merging the entry. The file contains a Sub2API API key, so keep it private and restrict its permissions where supported. Verify the effective configuration and make a smoke request:
+
+```bash
+grok inspect
+grok -p "Reply with sub2api-ok" -m grok
+```
+
+The `base_url` above is the public Sub2API URL ending in `/v1`, not `api.x.ai` or the internal xAI OAuth proxy URL.
 
 ### Usage And Quota Display
 
 xAI quota is passive. Sub2API does not invent subscription quota values; it records whitelisted xAI rate-limit headers from successful or rate-limited upstream responses when xAI sends them. Before the first usable upstream response, the dashboard shows quota as unknown and still displays local Sub2API usage stats.
 
-`401` responses mark the account as needing reauthorization. `403` responses are treated as entitlement or subscription-tier failures instead of token-refresh loops. `429` responses use `Retry-After` or a short cooldown to temporarily remove the account from scheduling.
+`401` responses temporarily remove accounts with invalid credentials from scheduling. `403` responses are treated as access or entitlement failures instead of token-refresh loops. `429` responses use `Retry-After` or a short cooldown to temporarily remove the account from scheduling.
 
 ---
 

README_CN.md

@@ -166,6 +166,11 @@
 </td>
 </tr>
 
+<tr>
+<td width="180"><a href="http://aimzoon.com"><img src="assets/partners/logos/aimzoon.jpg" alt="aimzoon" width="150"></a></td>
+<td>感谢 Aimzoon 对本项目的赞助！ <a href="http://aimzoon.com">Aimzoon</a> 提供稳定、高性价比的 AI API 接入服务，支持开发者将常用 AI 服务快速接入 Codex、Claude Code、Gemini CLI 等编程工具。无需复杂配置，更快接入，更稳调用，更省成本。codex倍率优惠，特价倍率等促销不断，注册即送免费体验额度，让 AI 编程真正进入日常工作流。<a href="http://aimzoon.com">点击这里</a>注册体验！
+</td>
+</tr>
 
 </table>
 
@@ -333,6 +338,7 @@ cd sub2api/deploy
 
 # 2. 复制环境配置文件
 cp .env.example .env
+chmod 600 .env
 
 # 3. 编辑配置（生成安全密码）
 nano .env
@@ -464,7 +470,23 @@ rm -rf data/ postgres_data/ redis_data/
 
 ---
 
-### 方式三：源码编译
+### 方式三：Apple container（macOS）
+
+Apple 芯片 Mac 在 macOS 26 上可使用 Apple `container` 1.1.0 或更高版本运行完整的 Sub2API、PostgreSQL 和 Redis：
+
+```bash
+git clone https://github.com/Wei-Shaw/sub2api.git
+cd sub2api/deploy
+./apple-container.sh init
+./apple-container.sh up
+./apple-container.sh status
+```
+
+该方式面向本地开发和人工运维，不提供持续重启监管；生产部署仍推荐 Docker Compose。生命周期命令、持久化、升级和运行时限制见 [deploy/APPLE_CONTAINER.md](deploy/APPLE_CONTAINER.md)。
+
+---
+
+### 方式四：源码编译
 
 从源码编译安装，适合开发或定制需求。
 

README_JA.md

@@ -161,6 +161,12 @@
 </td>
 </tr>
 
+<tr>
+<td width="180"><a href="http://aimzoon.com"><img src="assets/partners/logos/aimzoon.jpg" alt="aimzoon" width="150"></a></td>
+<td>Aimzoon のご支援に感謝します！<a href="http://aimzoon.com">Aimzoon</a> は安定してコストパフォーマンスに優れた AI API 接続サービスを提供し、開発者が主要な AI サービスを Codex、Claude Code、Gemini CLI などのコーディングツールへ素早く接続できるようにします。複雑な設定は不要で、より速い接続、より安定した呼び出し、より低いコストを実現。Codex レート割引や特価レートなどのキャンペーンも随時開催中、登録するだけで無料お試しクレジットをプレゼント。AI コーディングを日常のワークフローへ。<a href="http://aimzoon.com">こちら</a>から登録してお試しください！
+</td>
+</tr>
+
 </table>
 
 ## 概要
@@ -327,6 +333,7 @@ cd sub2api/deploy
 
 # 2. 環境設定ファイルをコピー
 cp .env.example .env
+chmod 600 .env
 
 # 3. 設定を編集（セキュアなパスワードを生成）
 nano .env
@@ -446,7 +453,23 @@ rm -rf data/ postgres_data/ redis_data/
 
 ---
 
-### 方法3: ソースからビルド
+### 方法3: Apple container（macOS）
+
+Apple シリコン搭載 Mac と macOS 26 では、Apple `container` 1.1.0 以降を使用して Sub2API、PostgreSQL、Redis の完全なスタックを実行できます:
+
+```bash
+git clone https://github.com/Wei-Shaw/sub2api.git
+cd sub2api/deploy
+./apple-container.sh init
+./apple-container.sh up
+./apple-container.sh status
+```
+
+これはローカル開発および手動運用向けです。本番環境では引き続き Docker Compose を推奨します。ライフサイクル、永続化、アップグレード、制限については [deploy/APPLE_CONTAINER.md](deploy/APPLE_CONTAINER.md) を参照してください。
+
+---
+
+### 方法4: ソースからビルド
 
 開発やカスタマイズのためにソースコードからビルドして実行します。
 

backend/cmd/jwtgen/main.go

@@ -48,7 +48,7 @@ func main() {
 		log.Fatalf("failed to resolve admin user: %v", err)
 	}
 
-	token, err := authService.GenerateToken(user)
+	token, err := authService.GenerateToken(ctx, user)
 	if err != nil {
 		log.Fatalf("failed to generate token: %v", err)
 	}

backend/cmd/server/VERSION

@@ -1 +1 @@
-0.1.150
+0.1.156

backend/cmd/server/wire.go

@@ -103,6 +103,8 @@ func provideCleanup(
 	paymentOrderExpiry *service.PaymentOrderExpiryService,
 	channelMonitorRunner *service.ChannelMonitorRunner,
 	quotaFlusher *service.UserPlatformQuotaUsageFlusher,
+	upstreamBillingProbe *service.UpstreamBillingProbeService,
+	auditLog *service.AuditLogService,
 ) func() {
 	return func() {
 		ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
@@ -133,6 +135,12 @@ func provideCleanup(
 				}
 				return nil
 			}},
+			{"AuditLogService", func() error {
+				if auditLog != nil {
+					auditLog.Stop()
+				}
+				return nil
+			}},
 			{"OpsAlertEvaluatorService", func() error {
 				if opsAlertEvaluator != nil {
 					opsAlertEvaluator.Stop()
@@ -279,6 +287,12 @@ func provideCleanup(
 				}
 				return nil
 			}},
+			{"UpstreamBillingProbeService", func() error {
+				if upstreamBillingProbe != nil {
+					upstreamBillingProbe.Stop()
+				}
+				return nil
+			}},
 		}
 
 		infraSteps := []cleanupStep{

backend/cmd/server/wire_gen.go

@@ -96,9 +96,6 @@ func initializeApplication(buildInfo handler.BuildInfo) (*Application, error) {
 	usageLogRepository := repository.NewUsageLogRepository(client, db)
 	usageService := service.NewUsageService(usageLogRepository, userRepository, client, apiKeyAuthCacheInvalidator)
 	opsRepository := repository.NewOpsRepository(db)
-	batchImageRepository := repository.NewBatchImageRepository(db)
-	batchImageQueue := repository.NewBatchImageQueue(redisClient, configConfig)
-	batchImageDownloadLimiter := repository.NewBatchImageDownloadLimiter(redisClient, configConfig)
 	usageBillingRepository := repository.NewUsageBillingRepository(client, db)
 	gatewayCache := repository.NewGatewayCache(redisClient)
 	schedulerOutboxRepository := repository.NewSchedulerOutboxRepository(db)
@@ -137,11 +134,6 @@ func initializeApplication(buildInfo handler.BuildInfo) (*Application, error) {
 	channelRepository := repository.NewChannelRepository(db)
 	channelService := service.NewChannelService(channelRepository, groupRepository, apiKeyAuthCacheInvalidator, pricingService)
 	modelPricingResolver := service.NewModelPricingResolver(channelService, billingService)
-	batchImageModelPricingResolver := service.ProvideBatchImageModelPricingResolver(modelPricingResolver)
-	batchImagePublicService := service.NewBatchImagePublicService(batchImageRepository, accountRepository, groupRepository, userGroupRateRepository, batchImageQueue, batchImageModelPricingResolver, usageBillingRepository, apiKeyAuthCacheInvalidator, configConfig)
-	batchImageDownloadService := service.NewBatchImageDownloadService(batchImageRepository, accountRepository, batchImageDownloadLimiter, configConfig)
-	batchImageCleanupService := service.ProvideBatchImageCleanupService(batchImageRepository, accountRepository, configConfig)
-	batchImageWorkerRuntime := service.ProvideBatchImageWorkerRuntime(batchImageRepository, accountRepository, batchImageQueue, usageBillingRepository, usageLogRepository, batchImageModelPricingResolver, apiKeyAuthCacheInvalidator, configConfig)
 	notificationEmailService := service.NewNotificationEmailService(settingRepository, emailService)
 	balanceNotifyService := service.ProvideBalanceNotifyService(emailService, settingRepository, accountRepository, notificationEmailService)
 	gatewayService := service.NewGatewayService(accountRepository, groupRepository, usageLogRepository, usageBillingRepository, userRepository, userSubscriptionRepository, userGroupRateRepository, gatewayCache, configConfig, schedulerSnapshotService, concurrencyService, billingService, rateLimitService, billingCacheService, identityService, httpUpstream, deferredService, claudeTokenProvider, sessionLimitCache, rpmCache, digestSessionStore, settingService, tlsFingerprintProfileService, channelService, modelPricingResolver, balanceNotifyService, serviceUserPlatformQuotaRepository)
@@ -181,21 +173,23 @@ func initializeApplication(buildInfo handler.BuildInfo) (*Application, error) {
 	leaderLockCache := repository.NewLeaderLockCache(redisClient)
 	dashboardAggregationService := service.ProvideDashboardAggregationService(dashboardAggregationRepository, timingWheelService, leaderLockCache, db, configConfig)
 	dashboardHandler := admin.NewDashboardHandler(dashboardService, dashboardAggregationService)
+	adminAccountRepository := repository.NewAdminAccountRepository(client, db, schedulerCache)
 	proxyExitInfoProber := repository.NewProxyExitInfoProber(configConfig)
 	proxyLatencyCache := repository.NewProxyLatencyCache(redisClient)
-	adminService := service.NewAdminService(userRepository, groupRepository, accountRepository, proxyRepository, apiKeyRepository, redeemCodeRepository, userGroupRateRepository, userRPMCache, billingCacheService, proxyExitInfoProber, proxyLatencyCache, apiKeyAuthCacheInvalidator, client, settingService, subscriptionService, userSubscriptionRepository, privacyClientFactory, openAIGatewayService)
-	adminUserHandler := admin.NewUserHandler(adminService, concurrencyService, serviceUserPlatformQuotaRepository, billingCache)
+	adminService := service.NewAdminService(userRepository, groupRepository, adminAccountRepository, proxyRepository, apiKeyRepository, redeemCodeRepository, userGroupRateRepository, userRPMCache, billingCacheService, proxyExitInfoProber, proxyLatencyCache, apiKeyAuthCacheInvalidator, client, settingService, subscriptionService, userSubscriptionRepository, privacyClientFactory, openAIGatewayService, affiliateService)
+	adminUserHandler := admin.NewUserHandler(adminService, concurrencyService, serviceUserPlatformQuotaRepository, billingCache, totpService, userService)
 	groupCapacityService := service.NewGroupCapacityService(accountRepository, groupRepository, concurrencyService, sessionLimitCache, rpmCache)
 	groupHandler := admin.NewGroupHandler(adminService, dashboardService, groupCapacityService)
 	claudeUsageFetcher := repository.NewClaudeUsageFetcher(httpUpstream)
 	antigravityQuotaFetcher := service.NewAntigravityQuotaFetcher(proxyRepository)
 	grokQuotaFetcher := service.NewGrokQuotaFetcher()
-	openAIQuotaService := service.ProvideOpenAIQuotaService(accountRepository, proxyRepository, openAITokenProvider, privacyClientFactory)
+	grokQuotaService := service.ProvideGrokQuotaService(accountRepository, proxyRepository, grokTokenProvider, httpUpstream, configConfig, usageLogRepository)
+	openAIQuotaService := service.ProvideOpenAIQuotaService(accountRepository, proxyRepository, openAITokenProvider, privacyClientFactory, openAIGatewayService)
 	usageCache := service.NewUsageCache()
-	accountUsageService := service.NewAccountUsageService(accountRepository, usageLogRepository, claudeUsageFetcher, geminiQuotaService, antigravityQuotaFetcher, grokQuotaFetcher, openAIQuotaService, usageCache, identityCache, tlsFingerprintProfileService)
-	accountTestService := service.NewAccountTestService(accountRepository, geminiTokenProvider, claudeTokenProvider, grokTokenProvider, antigravityGatewayService, httpUpstream, configConfig, tlsFingerprintProfileService)
+	accountUsageService := service.ProvideAccountUsageService(accountRepository, usageLogRepository, claudeUsageFetcher, geminiQuotaService, antigravityQuotaFetcher, grokQuotaFetcher, grokQuotaService, openAIQuotaService, usageCache, identityCache, tlsFingerprintProfileService, openAIGatewayService)
+	accountTestService := service.ProvideAccountTestService(accountRepository, geminiTokenProvider, claudeTokenProvider, grokTokenProvider, antigravityGatewayService, httpUpstream, configConfig, tlsFingerprintProfileService, openAIGatewayService)
 	crsSyncService := service.NewCRSSyncService(accountRepository, proxyRepository, oAuthService, openAIOAuthService, geminiOAuthService, configConfig)
-	accountHandler := admin.NewAccountHandler(adminService, oAuthService, openAIOAuthService, geminiOAuthService, antigravityOAuthService, rateLimitService, accountUsageService, accountTestService, concurrencyService, crsSyncService, sessionLimitCache, rpmCache, compositeTokenCacheInvalidator)
+	accountHandler := admin.ProvideAccountHandler(adminService, oAuthService, openAIOAuthService, geminiOAuthService, antigravityOAuthService, grokOAuthService, rateLimitService, accountUsageService, accountTestService, concurrencyService, crsSyncService, sessionLimitCache, rpmCache, compositeTokenCacheInvalidator, grokQuotaService)
 	adminAnnouncementHandler := admin.NewAnnouncementHandler(announcementService)
 	dataManagementService := service.NewDataManagementService()
 	dataManagementHandler := admin.NewDataManagementHandler(dataManagementService)
@@ -207,8 +201,8 @@ func initializeApplication(buildInfo handler.BuildInfo) (*Application, error) {
 	openAIOAuthHandler := admin.NewOpenAIOAuthHandler(openAIOAuthService, adminService, openAIQuotaService)
 	geminiOAuthHandler := admin.NewGeminiOAuthHandler(geminiOAuthService)
 	antigravityOAuthHandler := admin.NewAntigravityOAuthHandler(antigravityOAuthService)
-	grokQuotaService := service.ProvideGrokQuotaService(accountRepository, proxyRepository, grokTokenProvider, httpUpstream)
-	grokOAuthHandler := admin.NewGrokOAuthHandler(grokOAuthService, adminService, grokQuotaService)
+	tokenRefreshService := service.ProvideTokenRefreshService(accountRepository, oAuthService, openAIOAuthService, geminiOAuthService, antigravityOAuthService, grokOAuthService, compositeTokenCacheInvalidator, schedulerCache, configConfig, tempUnschedCache, privacyClientFactory, proxyRepository, oAuthRefreshAPI, openAIGatewayService)
+	grokOAuthHandler := admin.NewGrokOAuthHandler(grokOAuthService, adminService, grokQuotaService, tokenRefreshService)
 	proxyHandler := admin.NewProxyHandler(adminService)
 	adminRedeemHandler := admin.NewRedeemHandler(adminService, redeemService)
 	promoHandler := admin.NewPromoHandler(promoService)
@@ -256,40 +250,60 @@ func initializeApplication(buildInfo handler.BuildInfo) (*Application, error) {
 	paymentHandler := admin.NewPaymentHandler(paymentService, paymentConfigService)
 	affiliateHandler := admin.NewAffiliateHandler(affiliateService, adminService)
 	complianceHandler := admin.NewComplianceHandler(settingService)
-	adminHandlers := handler.ProvideAdminHandlers(dashboardHandler, adminUserHandler, groupHandler, accountHandler, adminAnnouncementHandler, dataManagementHandler, backupHandler, oAuthHandler, openAIOAuthHandler, geminiOAuthHandler, antigravityOAuthHandler, grokOAuthHandler, proxyHandler, adminRedeemHandler, promoHandler, settingHandler, opsHandler, systemHandler, adminSubscriptionHandler, adminUsageHandler, userAttributeHandler, errorPassthroughHandler, tlsFingerprintProfileHandler, adminAPIKeyHandler, scheduledTestHandler, channelHandler, channelMonitorHandler, channelMonitorRequestTemplateHandler, contentModerationHandler, paymentHandler, affiliateHandler, complianceHandler)
+	auditLogRepository := repository.NewAuditLogRepository(db)
+	auditLogService := service.ProvideAuditLogService(auditLogRepository, settingService)
+	auditLogHandler := admin.NewAuditLogHandler(auditLogService, totpService)
+	upstreamBillingProbeService := service.ProvideUpstreamBillingProbeService(accountRepository, accountTestService, settingService, leaderLockCache, db)
+	adminHandlers := handler.ProvideAdminHandlers(dashboardHandler, adminUserHandler, groupHandler, accountHandler, adminAnnouncementHandler, dataManagementHandler, backupHandler, oAuthHandler, openAIOAuthHandler, geminiOAuthHandler, antigravityOAuthHandler, grokOAuthHandler, proxyHandler, adminRedeemHandler, promoHandler, settingHandler, opsHandler, systemHandler, adminSubscriptionHandler, adminUsageHandler, userAttributeHandler, errorPassthroughHandler, tlsFingerprintProfileHandler, adminAPIKeyHandler, scheduledTestHandler, channelHandler, channelMonitorHandler, channelMonitorRequestTemplateHandler, contentModerationHandler, paymentHandler, affiliateHandler, complianceHandler, auditLogHandler, upstreamBillingProbeService)
 	usageRecordWorkerPool := service.NewUsageRecordWorkerPool(configConfig)
 	userMsgQueueCache := repository.NewUserMsgQueueCache(redisClient)
 	userMessageQueueService := service.ProvideUserMessageQueueService(userMsgQueueCache, rpmCache, configConfig)
-	gatewayHandler := handler.NewGatewayHandler(gatewayService, geminiMessagesCompatService, antigravityGatewayService, userService, concurrencyService, billingCacheService, usageService, apiKeyService, usageRecordWorkerPool, errorPassthroughService, contentModerationService, userMessageQueueService, configConfig, settingService)
+	gatewayHandler := handler.NewGatewayHandler(gatewayService, openAIGatewayService, geminiMessagesCompatService, antigravityGatewayService, userService, concurrencyService, billingCacheService, usageService, apiKeyService, usageRecordWorkerPool, errorPassthroughService, contentModerationService, userMessageQueueService, configConfig, settingService)
 	openAIGatewayHandler := handler.NewOpenAIGatewayHandler(openAIGatewayService, concurrencyService, billingCacheService, apiKeyService, usageRecordWorkerPool, errorPassthroughService, contentModerationService, opsService, configConfig)
 	handlerSettingHandler := handler.ProvideSettingHandler(settingService, buildInfo, notificationEmailService)
 	totpHandler := handler.NewTotpHandler(totpService)
-	handlerPaymentHandler := handler.NewPaymentHandler(paymentService, paymentConfigService, channelService)
+	handlerPaymentHandler := handler.NewPaymentHandler(paymentService, paymentConfigService)
 	paymentWebhookHandler := handler.NewPaymentWebhookHandler(paymentService, registry)
 	availableChannelHandler := handler.NewAvailableChannelHandler(channelService, apiKeyService, settingService)
+	imageTaskStore := repository.NewImageTaskStore(redisClient)
+	imageStorage, err := repository.ProvideImageStorage(configConfig)
+	if err != nil {
+		return nil, err
+	}
+	imageTaskService := service.ProvideImageTaskService(imageTaskStore, imageStorage, configConfig)
+	asyncImageHandler := handler.NewAsyncImageHandler(imageTaskService, openAIGatewayHandler)
+	batchImageRepository := repository.NewBatchImageRepository(db)
+	batchImageQueue := repository.NewBatchImageQueue(redisClient, configConfig)
+	batchImageModelPricingResolver := service.ProvideBatchImageModelPricingResolver(modelPricingResolver)
+	batchImagePublicService := service.NewBatchImagePublicService(batchImageRepository, accountRepository, groupRepository, userGroupRateRepository, batchImageQueue, batchImageModelPricingResolver, usageBillingRepository, apiKeyAuthCacheInvalidator, configConfig)
+	batchImageDownloadLimiter := repository.NewBatchImageDownloadLimiter(redisClient, configConfig)
+	batchImageDownloadService := service.NewBatchImageDownloadService(batchImageRepository, accountRepository, batchImageDownloadLimiter, configConfig)
+	batchImageCleanupService := service.ProvideBatchImageCleanupService(batchImageRepository, accountRepository, configConfig)
 	batchImageHandler := handler.NewBatchImageHandler(batchImagePublicService, batchImageDownloadService, batchImageCleanupService)
 	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, batchImageHandler, idempotencyCoordinator, idempotencyCleanupService)
-	jwtAuthMiddleware := middleware.NewJWTAuthMiddleware(authService, userService)
-	adminAuthMiddleware := middleware.NewAdminAuthMiddleware(authService, userService, settingService)
+	handlers := handler.ProvideHandlers(authHandler, userHandler, apiKeyHandler, usageHandler, redeemHandler, subscriptionHandler, announcementHandler, channelMonitorUserHandler, adminHandlers, gatewayHandler, openAIGatewayHandler, handlerSettingHandler, totpHandler, handlerPaymentHandler, paymentWebhookHandler, availableChannelHandler, asyncImageHandler, batchImageHandler, idempotencyCoordinator, idempotencyCleanupService)
+	jwtAuthMiddleware := middleware.NewJWTAuthMiddleware(authService, userService, settingService, auditLogService)
+	adminAuthMiddleware := middleware.NewAdminAuthMiddleware(authService, userService, settingService, auditLogService)
 	apiKeyAuthMiddleware := middleware.NewAPIKeyAuthMiddleware(apiKeyService, subscriptionService, configConfig)
-	engine := server.ProvideRouter(configConfig, handlers, jwtAuthMiddleware, adminAuthMiddleware, apiKeyAuthMiddleware, apiKeyService, subscriptionService, opsService, settingService, redisClient)
+	auditLogMiddleware := middleware.NewAuditLogMiddleware(auditLogService)
+	stepUpAuthMiddleware := middleware.NewStepUpAuthMiddleware(totpService, userService)
+	engine := server.ProvideRouter(configConfig, handlers, jwtAuthMiddleware, adminAuthMiddleware, apiKeyAuthMiddleware, auditLogMiddleware, stepUpAuthMiddleware, apiKeyService, subscriptionService, opsService, settingService, redisClient)
 	httpServer := server.ProvideHTTPServer(configConfig, engine)
 	opsMetricsCollector := service.ProvideOpsMetricsCollector(opsRepository, settingRepository, accountRepository, concurrencyService, db, redisClient, configConfig)
 	opsAggregationService := service.ProvideOpsAggregationService(opsRepository, settingRepository, db, redisClient, configConfig)
 	opsAlertEvaluatorService := service.ProvideOpsAlertEvaluatorService(opsService, opsRepository, emailService, redisClient, configConfig, proxyRepository)
 	opsCleanupService := service.ProvideOpsCleanupService(opsRepository, db, redisClient, configConfig, channelMonitorService, settingRepository, opsService)
 	opsScheduledReportService := service.ProvideOpsScheduledReportService(opsService, userService, emailService, redisClient, configConfig)
-	tokenRefreshService := service.ProvideTokenRefreshService(accountRepository, oAuthService, openAIOAuthService, geminiOAuthService, antigravityOAuthService, grokOAuthService, compositeTokenCacheInvalidator, schedulerCache, configConfig, tempUnschedCache, privacyClientFactory, proxyRepository, oAuthRefreshAPI, openAIGatewayService)
 	accountExpiryService := service.ProvideAccountExpiryService(accountRepository)
 	proxyExpiryService := service.ProvideProxyExpiryService(proxyRepository)
 	subscriptionExpiryService := service.ProvideSubscriptionExpiryService(userSubscriptionRepository, settingRepository, notificationEmailService, leaderLockCache, db)
+	batchImageWorkerRuntime := service.ProvideBatchImageWorkerRuntime(batchImageRepository, accountRepository, batchImageQueue, usageBillingRepository, usageLogRepository, batchImageModelPricingResolver, apiKeyAuthCacheInvalidator, configConfig)
 	scheduledTestRunnerService := service.ProvideScheduledTestRunnerService(scheduledTestPlanRepository, scheduledTestService, accountTestService, rateLimitService, configConfig)
 	paymentOrderExpiryService := service.ProvidePaymentOrderExpiryService(paymentService, leaderLockCache, db)
 	channelMonitorRunner := service.ProvideChannelMonitorRunner(channelMonitorService, settingService)
 	userPlatformQuotaUsageFlusher := service.ProvideUserPlatformQuotaUsageFlusher(configConfig, billingCache, serviceUserPlatformQuotaRepository, timingWheelService)
-	v := provideCleanup(client, redisClient, opsMetricsCollector, opsAggregationService, opsAlertEvaluatorService, opsCleanupService, opsScheduledReportService, opsSystemLogSink, schedulerSnapshotService, tokenRefreshService, accountExpiryService, proxyExpiryService, subscriptionExpiryService, usageCleanupService, idempotencyCleanupService, batchImageCleanupService, batchImageWorkerRuntime, pricingService, emailQueueService, billingCacheService, usageRecordWorkerPool, subscriptionService, oAuthService, openAIOAuthService, geminiOAuthService, antigravityOAuthService, grokOAuthService, openAIGatewayService, scheduledTestRunnerService, backupService, paymentOrderExpiryService, channelMonitorRunner, userPlatformQuotaUsageFlusher)
+	v := provideCleanup(client, redisClient, opsMetricsCollector, opsAggregationService, opsAlertEvaluatorService, opsCleanupService, opsScheduledReportService, opsSystemLogSink, schedulerSnapshotService, tokenRefreshService, accountExpiryService, proxyExpiryService, subscriptionExpiryService, usageCleanupService, idempotencyCleanupService, batchImageCleanupService, batchImageWorkerRuntime, pricingService, emailQueueService, billingCacheService, usageRecordWorkerPool, subscriptionService, oAuthService, openAIOAuthService, geminiOAuthService, antigravityOAuthService, grokOAuthService, openAIGatewayService, scheduledTestRunnerService, backupService, paymentOrderExpiryService, channelMonitorRunner, userPlatformQuotaUsageFlusher, upstreamBillingProbeService, auditLogService)
 	application := &Application{
 		Server:  httpServer,
 		Cleanup: v,
@@ -349,6 +363,8 @@ func provideCleanup(
 	paymentOrderExpiry *service.PaymentOrderExpiryService,
 	channelMonitorRunner *service.ChannelMonitorRunner,
 	quotaFlusher *service.UserPlatformQuotaUsageFlusher,
+	upstreamBillingProbe *service.UpstreamBillingProbeService,
+	auditLog *service.AuditLogService,
 ) func() {
 	return func() {
 		ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
@@ -378,6 +394,12 @@ func provideCleanup(
 				}
 				return nil
 			}},
+			{"AuditLogService", func() error {
+				if auditLog != nil {
+					auditLog.Stop()
+				}
+				return nil
+			}},
 			{"OpsAlertEvaluatorService", func() error {
 				if opsAlertEvaluator != nil {
 					opsAlertEvaluator.Stop()
@@ -524,6 +546,12 @@ func provideCleanup(
 				}
 				return nil
 			}},
+			{"UpstreamBillingProbeService", func() error {
+				if upstreamBillingProbe != nil {
+					upstreamBillingProbe.Stop()
+				}
+				return nil
+			}},
 		}
 
 		infraSteps := []cleanupStep{

backend/cmd/server/wire_gen_test.go

@@ -83,6 +83,8 @@ func TestProvideCleanup_WithMinimalDependencies_NoPanic(t *testing.T) {
 		nil, // paymentOrderExpiry
 		nil, // channelMonitorRunner
 		nil, // quotaFlusher
+		nil, // upstreamBillingProbe
+		nil, // auditLog
 	)
 
 	require.NotPanics(t, func() {

backend/ent/channelmonitor/channelmonitor.go

@@ -167,6 +167,7 @@ const (
 	ProviderOpenai    Provider = "openai"
 	ProviderAnthropic Provider = "anthropic"
 	ProviderGemini    Provider = "gemini"
+	ProviderGrok      Provider = "grok"
 )
 
 func (pr Provider) String() string {
@@ -176,7 +177,7 @@ func (pr Provider) String() string {
 // ProviderValidator is a validator for the "provider" field enum values. It is called by the builders before save.
 func ProviderValidator(pr Provider) error {
 	switch pr {
-	case ProviderOpenai, ProviderAnthropic, ProviderGemini:
+	case ProviderOpenai, ProviderAnthropic, ProviderGemini, ProviderGrok:
 		return nil
 	default:
 		return fmt.Errorf("channelmonitor: invalid enum value for provider field: %q", pr)

backend/ent/channelmonitorrequesttemplate/channelmonitorrequesttemplate.go

@@ -103,6 +103,7 @@ const (
 	ProviderOpenai    Provider = "openai"
 	ProviderAnthropic Provider = "anthropic"
 	ProviderGemini    Provider = "gemini"
+	ProviderGrok      Provider = "grok"
 )
 
 func (pr Provider) String() string {
@@ -112,7 +113,7 @@ func (pr Provider) String() string {
 // ProviderValidator is a validator for the "provider" field enum values. It is called by the builders before save.
 func ProviderValidator(pr Provider) error {
 	switch pr {
-	case ProviderOpenai, ProviderAnthropic, ProviderGemini:
+	case ProviderOpenai, ProviderAnthropic, ProviderGemini, ProviderGrok:
 		return nil
 	default:
 		return fmt.Errorf("channelmonitorrequesttemplate: invalid enum value for provider field: %q", pr)

backend/ent/group.go

@@ -83,6 +83,8 @@ type Group struct {
 	VideoPrice720p *float64 `json:"video_price_720p,omitempty"`
 	// VideoPrice1080p holds the value of the "video_price_1080p" field.
 	VideoPrice1080p *float64 `json:"video_price_1080p,omitempty"`
+	// Codex alpha/search 网页搜索单次价格（USD/次）；nil 表示使用默认价 0.01（官方 $10/1000 次）
+	WebSearchPricePerCall *float64 `json:"web_search_price_per_call,omitempty"`
 	// 是否仅允许 Claude Code 客户端
 	ClaudeCodeOnly bool `json:"claude_code_only,omitempty"`
 	// 非 Claude Code 请求降级使用的分组 ID
@@ -223,7 +225,7 @@ func (*Group) scanValues(columns []string) ([]any, error) {
 			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:
 			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:
+		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:
 			values[i] = new(sql.NullFloat64)
 		case group.FieldID, group.FieldDefaultValidityDays, group.FieldFallbackGroupID, group.FieldFallbackGroupIDOnInvalidRequest, group.FieldSortOrder, group.FieldRpmLimit:
 			values[i] = new(sql.NullInt64)
@@ -455,6 +457,13 @@ func (_m *Group) assignValues(columns []string, values []any) error {
 				_m.VideoPrice1080p = new(float64)
 				*_m.VideoPrice1080p = value.Float64
 			}
+		case group.FieldWebSearchPricePerCall:
+			if value, ok := values[i].(*sql.NullFloat64); !ok {
+				return fmt.Errorf("unexpected type %T for field web_search_price_per_call", values[i])
+			} else if value.Valid {
+				_m.WebSearchPricePerCall = new(float64)
+				*_m.WebSearchPricePerCall = value.Float64
+			}
 		case group.FieldClaudeCodeOnly:
 			if value, ok := values[i].(*sql.NullBool); !ok {
 				return fmt.Errorf("unexpected type %T for field claude_code_only", values[i])
@@ -749,6 +758,11 @@ func (_m *Group) String() string {
 		builder.WriteString(fmt.Sprintf("%v", *v))
 	}
 	builder.WriteString(", ")
+	if v := _m.WebSearchPricePerCall; v != nil {
+		builder.WriteString("web_search_price_per_call=")
+		builder.WriteString(fmt.Sprintf("%v", *v))
+	}
+	builder.WriteString(", ")
 	builder.WriteString("claude_code_only=")
 	builder.WriteString(fmt.Sprintf("%v", _m.ClaudeCodeOnly))
 	builder.WriteString(", ")

backend/ent/group/group.go

@@ -80,6 +80,8 @@ const (
 	FieldVideoPrice720p = "video_price_720p"
 	// FieldVideoPrice1080p holds the string denoting the video_price_1080p field in the database.
 	FieldVideoPrice1080p = "video_price_1080p"
+	// FieldWebSearchPricePerCall holds the string denoting the web_search_price_per_call field in the database.
+	FieldWebSearchPricePerCall = "web_search_price_per_call"
 	// FieldClaudeCodeOnly holds the string denoting the claude_code_only field in the database.
 	FieldClaudeCodeOnly = "claude_code_only"
 	// FieldFallbackGroupID holds the string denoting the fallback_group_id field in the database.
@@ -217,6 +219,7 @@ var Columns = []string{
 	FieldVideoPrice480p,
 	FieldVideoPrice720p,
 	FieldVideoPrice1080p,
+	FieldWebSearchPricePerCall,
 	FieldClaudeCodeOnly,
 	FieldFallbackGroupID,
 	FieldFallbackGroupIDOnInvalidRequest,
@@ -511,6 +514,11 @@ func ByVideoPrice1080p(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldVideoPrice1080p, opts...).ToFunc()
 }
 
+// ByWebSearchPricePerCall orders the results by the web_search_price_per_call field.
+func ByWebSearchPricePerCall(opts ...sql.OrderTermOption) OrderOption {
+	return sql.OrderByField(FieldWebSearchPricePerCall, opts...).ToFunc()
+}
+
 // ByClaudeCodeOnly orders the results by the claude_code_only field.
 func ByClaudeCodeOnly(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldClaudeCodeOnly, opts...).ToFunc()

backend/ent/group/where.go

@@ -215,6 +215,11 @@ func VideoPrice1080p(v float64) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldVideoPrice1080p, v))
 }
 
+// WebSearchPricePerCall applies equality check predicate on the "web_search_price_per_call" field. It's identical to WebSearchPricePerCallEQ.
+func WebSearchPricePerCall(v float64) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldWebSearchPricePerCall, v))
+}
+
 // ClaudeCodeOnly applies equality check predicate on the "claude_code_only" field. It's identical to ClaudeCodeOnlyEQ.
 func ClaudeCodeOnly(v bool) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldClaudeCodeOnly, v))
@@ -1655,6 +1660,56 @@ func VideoPrice1080pNotNil() predicate.Group {
 	return predicate.Group(sql.FieldNotNull(FieldVideoPrice1080p))
 }
 
+// WebSearchPricePerCallEQ applies the EQ predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallEQ(v float64) predicate.Group {
+	return predicate.Group(sql.FieldEQ(FieldWebSearchPricePerCall, v))
+}
+
+// WebSearchPricePerCallNEQ applies the NEQ predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallNEQ(v float64) predicate.Group {
+	return predicate.Group(sql.FieldNEQ(FieldWebSearchPricePerCall, v))
+}
+
+// WebSearchPricePerCallIn applies the In predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallIn(vs ...float64) predicate.Group {
+	return predicate.Group(sql.FieldIn(FieldWebSearchPricePerCall, vs...))
+}
+
+// WebSearchPricePerCallNotIn applies the NotIn predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallNotIn(vs ...float64) predicate.Group {
+	return predicate.Group(sql.FieldNotIn(FieldWebSearchPricePerCall, vs...))
+}
+
+// WebSearchPricePerCallGT applies the GT predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallGT(v float64) predicate.Group {
+	return predicate.Group(sql.FieldGT(FieldWebSearchPricePerCall, v))
+}
+
+// WebSearchPricePerCallGTE applies the GTE predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallGTE(v float64) predicate.Group {
+	return predicate.Group(sql.FieldGTE(FieldWebSearchPricePerCall, v))
+}
+
+// WebSearchPricePerCallLT applies the LT predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallLT(v float64) predicate.Group {
+	return predicate.Group(sql.FieldLT(FieldWebSearchPricePerCall, v))
+}
+
+// WebSearchPricePerCallLTE applies the LTE predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallLTE(v float64) predicate.Group {
+	return predicate.Group(sql.FieldLTE(FieldWebSearchPricePerCall, v))
+}
+
+// WebSearchPricePerCallIsNil applies the IsNil predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallIsNil() predicate.Group {
+	return predicate.Group(sql.FieldIsNull(FieldWebSearchPricePerCall))
+}
+
+// WebSearchPricePerCallNotNil applies the NotNil predicate on the "web_search_price_per_call" field.
+func WebSearchPricePerCallNotNil() predicate.Group {
+	return predicate.Group(sql.FieldNotNull(FieldWebSearchPricePerCall))
+}
+
 // ClaudeCodeOnlyEQ applies the EQ predicate on the "claude_code_only" field.
 func ClaudeCodeOnlyEQ(v bool) predicate.Group {
 	return predicate.Group(sql.FieldEQ(FieldClaudeCodeOnly, v))

backend/ent/group_create.go

@@ -469,6 +469,20 @@ func (_c *GroupCreate) SetNillableVideoPrice1080p(v *float64) *GroupCreate {
 	return _c
 }
 
+// SetWebSearchPricePerCall sets the "web_search_price_per_call" field.
+func (_c *GroupCreate) SetWebSearchPricePerCall(v float64) *GroupCreate {
+	_c.mutation.SetWebSearchPricePerCall(v)
+	return _c
+}
+
+// SetNillableWebSearchPricePerCall sets the "web_search_price_per_call" field if the given value is not nil.
+func (_c *GroupCreate) SetNillableWebSearchPricePerCall(v *float64) *GroupCreate {
+	if v != nil {
+		_c.SetWebSearchPricePerCall(*v)
+	}
+	return _c
+}
+
 // SetClaudeCodeOnly sets the "claude_code_only" field.
 func (_c *GroupCreate) SetClaudeCodeOnly(v bool) *GroupCreate {
 	_c.mutation.SetClaudeCodeOnly(v)
@@ -1218,6 +1232,10 @@ func (_c *GroupCreate) createSpec() (*Group, *sqlgraph.CreateSpec) {
 		_spec.SetField(group.FieldVideoPrice1080p, field.TypeFloat64, value)
 		_node.VideoPrice1080p = &value
 	}
+	if value, ok := _c.mutation.WebSearchPricePerCall(); ok {
+		_spec.SetField(group.FieldWebSearchPricePerCall, field.TypeFloat64, value)
+		_node.WebSearchPricePerCall = &value
+	}
 	if value, ok := _c.mutation.ClaudeCodeOnly(); ok {
 		_spec.SetField(group.FieldClaudeCodeOnly, field.TypeBool, value)
 		_node.ClaudeCodeOnly = value
@@ -1968,6 +1986,30 @@ func (u *GroupUpsert) ClearVideoPrice1080p() *GroupUpsert {
 	return u
 }
 
+// SetWebSearchPricePerCall sets the "web_search_price_per_call" field.
+func (u *GroupUpsert) SetWebSearchPricePerCall(v float64) *GroupUpsert {
+	u.Set(group.FieldWebSearchPricePerCall, v)
+	return u
+}
+
+// UpdateWebSearchPricePerCall sets the "web_search_price_per_call" field to the value that was provided on create.
+func (u *GroupUpsert) UpdateWebSearchPricePerCall() *GroupUpsert {
+	u.SetExcluded(group.FieldWebSearchPricePerCall)
+	return u
+}
+
+// AddWebSearchPricePerCall adds v to the "web_search_price_per_call" field.
+func (u *GroupUpsert) AddWebSearchPricePerCall(v float64) *GroupUpsert {
+	u.Add(group.FieldWebSearchPricePerCall, v)
+	return u
+}
+
+// ClearWebSearchPricePerCall clears the value of the "web_search_price_per_call" field.
+func (u *GroupUpsert) ClearWebSearchPricePerCall() *GroupUpsert {
+	u.SetNull(group.FieldWebSearchPricePerCall)
+	return u
+}
+
 // SetClaudeCodeOnly sets the "claude_code_only" field.
 func (u *GroupUpsert) SetClaudeCodeOnly(v bool) *GroupUpsert {
 	u.Set(group.FieldClaudeCodeOnly, v)
@@ -2858,6 +2900,34 @@ func (u *GroupUpsertOne) ClearVideoPrice1080p() *GroupUpsertOne {
 	})
 }
 
+// SetWebSearchPricePerCall sets the "web_search_price_per_call" field.
+func (u *GroupUpsertOne) SetWebSearchPricePerCall(v float64) *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetWebSearchPricePerCall(v)
+	})
+}
+
+// AddWebSearchPricePerCall adds v to the "web_search_price_per_call" field.
+func (u *GroupUpsertOne) AddWebSearchPricePerCall(v float64) *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.AddWebSearchPricePerCall(v)
+	})
+}
+
+// UpdateWebSearchPricePerCall sets the "web_search_price_per_call" field to the value that was provided on create.
+func (u *GroupUpsertOne) UpdateWebSearchPricePerCall() *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateWebSearchPricePerCall()
+	})
+}
+
+// ClearWebSearchPricePerCall clears the value of the "web_search_price_per_call" field.
+func (u *GroupUpsertOne) ClearWebSearchPricePerCall() *GroupUpsertOne {
+	return u.Update(func(s *GroupUpsert) {
+		s.ClearWebSearchPricePerCall()
+	})
+}
+
 // SetClaudeCodeOnly sets the "claude_code_only" field.
 func (u *GroupUpsertOne) SetClaudeCodeOnly(v bool) *GroupUpsertOne {
 	return u.Update(func(s *GroupUpsert) {
@@ -3951,6 +4021,34 @@ func (u *GroupUpsertBulk) ClearVideoPrice1080p() *GroupUpsertBulk {
 	})
 }
 
+// SetWebSearchPricePerCall sets the "web_search_price_per_call" field.
+func (u *GroupUpsertBulk) SetWebSearchPricePerCall(v float64) *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.SetWebSearchPricePerCall(v)
+	})
+}
+
+// AddWebSearchPricePerCall adds v to the "web_search_price_per_call" field.
+func (u *GroupUpsertBulk) AddWebSearchPricePerCall(v float64) *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.AddWebSearchPricePerCall(v)
+	})
+}
+
+// UpdateWebSearchPricePerCall sets the "web_search_price_per_call" field to the value that was provided on create.
+func (u *GroupUpsertBulk) UpdateWebSearchPricePerCall() *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.UpdateWebSearchPricePerCall()
+	})
+}
+
+// ClearWebSearchPricePerCall clears the value of the "web_search_price_per_call" field.
+func (u *GroupUpsertBulk) ClearWebSearchPricePerCall() *GroupUpsertBulk {
+	return u.Update(func(s *GroupUpsert) {
+		s.ClearWebSearchPricePerCall()
+	})
+}
+
 // SetClaudeCodeOnly sets the "claude_code_only" field.
 func (u *GroupUpsertBulk) SetClaudeCodeOnly(v bool) *GroupUpsertBulk {
 	return u.Update(func(s *GroupUpsert) {

backend/ent/group_update.go

@@ -640,6 +640,33 @@ func (_u *GroupUpdate) ClearVideoPrice1080p() *GroupUpdate {
 	return _u
 }
 
+// SetWebSearchPricePerCall sets the "web_search_price_per_call" field.
+func (_u *GroupUpdate) SetWebSearchPricePerCall(v float64) *GroupUpdate {
+	_u.mutation.ResetWebSearchPricePerCall()
+	_u.mutation.SetWebSearchPricePerCall(v)
+	return _u
+}
+
+// SetNillableWebSearchPricePerCall sets the "web_search_price_per_call" field if the given value is not nil.
+func (_u *GroupUpdate) SetNillableWebSearchPricePerCall(v *float64) *GroupUpdate {
+	if v != nil {
+		_u.SetWebSearchPricePerCall(*v)
+	}
+	return _u
+}
+
+// AddWebSearchPricePerCall adds value to the "web_search_price_per_call" field.
+func (_u *GroupUpdate) AddWebSearchPricePerCall(v float64) *GroupUpdate {
+	_u.mutation.AddWebSearchPricePerCall(v)
+	return _u
+}
+
+// ClearWebSearchPricePerCall clears the value of the "web_search_price_per_call" field.
+func (_u *GroupUpdate) ClearWebSearchPricePerCall() *GroupUpdate {
+	_u.mutation.ClearWebSearchPricePerCall()
+	return _u
+}
+
 // SetClaudeCodeOnly sets the "claude_code_only" field.
 func (_u *GroupUpdate) SetClaudeCodeOnly(v bool) *GroupUpdate {
 	_u.mutation.SetClaudeCodeOnly(v)
@@ -1375,6 +1402,15 @@ func (_u *GroupUpdate) sqlSave(ctx context.Context) (_node int, err error) {
 	if _u.mutation.VideoPrice1080pCleared() {
 		_spec.ClearField(group.FieldVideoPrice1080p, field.TypeFloat64)
 	}
+	if value, ok := _u.mutation.WebSearchPricePerCall(); ok {
+		_spec.SetField(group.FieldWebSearchPricePerCall, field.TypeFloat64, value)
+	}
+	if value, ok := _u.mutation.AddedWebSearchPricePerCall(); ok {
+		_spec.AddField(group.FieldWebSearchPricePerCall, field.TypeFloat64, value)
+	}
+	if _u.mutation.WebSearchPricePerCallCleared() {
+		_spec.ClearField(group.FieldWebSearchPricePerCall, field.TypeFloat64)
+	}
 	if value, ok := _u.mutation.ClaudeCodeOnly(); ok {
 		_spec.SetField(group.FieldClaudeCodeOnly, field.TypeBool, value)
 	}
@@ -2364,6 +2400,33 @@ func (_u *GroupUpdateOne) ClearVideoPrice1080p() *GroupUpdateOne {
 	return _u
 }
 
+// SetWebSearchPricePerCall sets the "web_search_price_per_call" field.
+func (_u *GroupUpdateOne) SetWebSearchPricePerCall(v float64) *GroupUpdateOne {
+	_u.mutation.ResetWebSearchPricePerCall()
+	_u.mutation.SetWebSearchPricePerCall(v)
+	return _u
+}
+
+// SetNillableWebSearchPricePerCall sets the "web_search_price_per_call" field if the given value is not nil.
+func (_u *GroupUpdateOne) SetNillableWebSearchPricePerCall(v *float64) *GroupUpdateOne {
+	if v != nil {
+		_u.SetWebSearchPricePerCall(*v)
+	}
+	return _u
+}
+
+// AddWebSearchPricePerCall adds value to the "web_search_price_per_call" field.
+func (_u *GroupUpdateOne) AddWebSearchPricePerCall(v float64) *GroupUpdateOne {
+	_u.mutation.AddWebSearchPricePerCall(v)
+	return _u
+}
+
+// ClearWebSearchPricePerCall clears the value of the "web_search_price_per_call" field.
+func (_u *GroupUpdateOne) ClearWebSearchPricePerCall() *GroupUpdateOne {
+	_u.mutation.ClearWebSearchPricePerCall()
+	return _u
+}
+
 // SetClaudeCodeOnly sets the "claude_code_only" field.
 func (_u *GroupUpdateOne) SetClaudeCodeOnly(v bool) *GroupUpdateOne {
 	_u.mutation.SetClaudeCodeOnly(v)
@@ -3129,6 +3192,15 @@ func (_u *GroupUpdateOne) sqlSave(ctx context.Context) (_node *Group, err error)
 	if _u.mutation.VideoPrice1080pCleared() {
 		_spec.ClearField(group.FieldVideoPrice1080p, field.TypeFloat64)
 	}
+	if value, ok := _u.mutation.WebSearchPricePerCall(); ok {
+		_spec.SetField(group.FieldWebSearchPricePerCall, field.TypeFloat64, value)
+	}
+	if value, ok := _u.mutation.AddedWebSearchPricePerCall(); ok {
+		_spec.AddField(group.FieldWebSearchPricePerCall, field.TypeFloat64, value)
+	}
+	if _u.mutation.WebSearchPricePerCallCleared() {
+		_spec.ClearField(group.FieldWebSearchPricePerCall, field.TypeFloat64)
+	}
 	if value, ok := _u.mutation.ClaudeCodeOnly(); ok {
 		_spec.SetField(group.FieldClaudeCodeOnly, field.TypeBool, value)
 	}

backend/ent/migrate/schema.go

@@ -623,7 +623,7 @@ var (
 		{Name: "created_at", Type: field.TypeTime, SchemaType: map[string]string{"postgres": "timestamptz"}},
 		{Name: "updated_at", Type: field.TypeTime, SchemaType: map[string]string{"postgres": "timestamptz"}},
 		{Name: "name", Type: field.TypeString, Size: 100},
-		{Name: "provider", Type: field.TypeEnum, Enums: []string{"openai", "anthropic", "gemini"}},
+		{Name: "provider", Type: field.TypeEnum, Enums: []string{"openai", "anthropic", "gemini", "grok"}},
 		{Name: "api_mode", Type: field.TypeString, Size: 32, Default: "chat_completions"},
 		{Name: "endpoint", Type: field.TypeString, Size: 500},
 		{Name: "api_key_encrypted", Type: field.TypeString},
@@ -768,7 +768,7 @@ var (
 		{Name: "created_at", Type: field.TypeTime, SchemaType: map[string]string{"postgres": "timestamptz"}},
 		{Name: "updated_at", Type: field.TypeTime, SchemaType: map[string]string{"postgres": "timestamptz"}},
 		{Name: "name", Type: field.TypeString, Size: 100},
-		{Name: "provider", Type: field.TypeEnum, Enums: []string{"openai", "anthropic", "gemini"}},
+		{Name: "provider", Type: field.TypeEnum, Enums: []string{"openai", "anthropic", "gemini", "grok"}},
 		{Name: "api_mode", Type: field.TypeString, Size: 32, Default: "chat_completions"},
 		{Name: "description", Type: field.TypeString, Nullable: true, Size: 500, Default: ""},
 		{Name: "extra_headers", Type: field.TypeJSON},
@@ -865,6 +865,7 @@ var (
 		{Name: "video_price_480p", Type: field.TypeFloat64, Nullable: true, SchemaType: map[string]string{"postgres": "decimal(20,8)"}},
 		{Name: "video_price_720p", Type: field.TypeFloat64, Nullable: true, SchemaType: map[string]string{"postgres": "decimal(20,8)"}},
 		{Name: "video_price_1080p", Type: field.TypeFloat64, Nullable: true, SchemaType: map[string]string{"postgres": "decimal(20,8)"}},
+		{Name: "web_search_price_per_call", Type: field.TypeFloat64, Nullable: true, SchemaType: map[string]string{"postgres": "decimal(20,8)"}},
 		{Name: "claude_code_only", Type: field.TypeBool, Default: false},
 		{Name: "fallback_group_id", Type: field.TypeInt64, Nullable: true},
 		{Name: "fallback_group_id_on_invalid_request", Type: field.TypeInt64, Nullable: true},
@@ -915,7 +916,7 @@ var (
 			{
 				Name:    "group_sort_order",
 				Unique:  false,
-				Columns: []*schema.Column{GroupsColumns[40]},
+				Columns: []*schema.Column{GroupsColumns[41]},
 			},
 		},
 	}
@@ -1446,6 +1447,7 @@ var (
 		{Name: "description", Type: field.TypeString, Default: "", SchemaType: map[string]string{"postgres": "text"}},
 		{Name: "price", Type: field.TypeFloat64, SchemaType: map[string]string{"postgres": "decimal(20,2)"}},
 		{Name: "original_price", Type: field.TypeFloat64, Nullable: true, SchemaType: map[string]string{"postgres": "decimal(20,2)"}},
+		{Name: "currency", Type: field.TypeString, Size: 3, Default: ""},
 		{Name: "validity_days", Type: field.TypeInt, Default: 30},
 		{Name: "validity_unit", Type: field.TypeString, Size: 10, Default: "day"},
 		{Name: "features", Type: field.TypeString, Default: "", SchemaType: map[string]string{"postgres": "text"}},
@@ -1469,7 +1471,7 @@ var (
 			{
 				Name:    "subscriptionplan_for_sale",
 				Unique:  false,
-				Columns: []*schema.Column{SubscriptionPlansColumns[10]},
+				Columns: []*schema.Column{SubscriptionPlansColumns[11]},
 			},
 		},
 	}
@@ -1559,6 +1561,7 @@ var (
 		{Name: "total_cost", Type: field.TypeFloat64, Default: 0, SchemaType: map[string]string{"postgres": "decimal(20,10)"}},
 		{Name: "actual_cost", Type: field.TypeFloat64, Default: 0, SchemaType: map[string]string{"postgres": "decimal(20,10)"}},
 		{Name: "rate_multiplier", Type: field.TypeFloat64, Default: 1, SchemaType: map[string]string{"postgres": "decimal(10,4)"}},
+		{Name: "long_context_billing_applied", Type: field.TypeBool, Default: false},
 		{Name: "account_rate_multiplier", Type: field.TypeFloat64, Nullable: true, SchemaType: map[string]string{"postgres": "decimal(10,4)"}},
 		{Name: "billing_type", Type: field.TypeInt8, Default: 0},
 		{Name: "stream", Type: field.TypeBool, Default: false},
@@ -1591,31 +1594,31 @@ var (
 		ForeignKeys: []*schema.ForeignKey{
 			{
 				Symbol:     "usage_logs_api_keys_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[40]},
+				Columns:    []*schema.Column{UsageLogsColumns[41]},
 				RefColumns: []*schema.Column{APIKeysColumns[0]},
 				OnDelete:   schema.NoAction,
 			},
 			{
 				Symbol:     "usage_logs_accounts_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[41]},
+				Columns:    []*schema.Column{UsageLogsColumns[42]},
 				RefColumns: []*schema.Column{AccountsColumns[0]},
 				OnDelete:   schema.NoAction,
 			},
 			{
 				Symbol:     "usage_logs_groups_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[42]},
+				Columns:    []*schema.Column{UsageLogsColumns[43]},
 				RefColumns: []*schema.Column{GroupsColumns[0]},
 				OnDelete:   schema.SetNull,
 			},
 			{
 				Symbol:     "usage_logs_users_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[43]},
+				Columns:    []*schema.Column{UsageLogsColumns[44]},
 				RefColumns: []*schema.Column{UsersColumns[0]},
 				OnDelete:   schema.NoAction,
 			},
 			{
 				Symbol:     "usage_logs_user_subscriptions_usage_logs",
-				Columns:    []*schema.Column{UsageLogsColumns[44]},
+				Columns:    []*schema.Column{UsageLogsColumns[45]},
 				RefColumns: []*schema.Column{UserSubscriptionsColumns[0]},
 				OnDelete:   schema.SetNull,
 			},
@@ -1624,32 +1627,32 @@ var (
 			{
 				Name:    "usagelog_user_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[43]},
+				Columns: []*schema.Column{UsageLogsColumns[44]},
 			},
 			{
 				Name:    "usagelog_api_key_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[40]},
+				Columns: []*schema.Column{UsageLogsColumns[41]},
 			},
 			{
 				Name:    "usagelog_account_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[41]},
+				Columns: []*schema.Column{UsageLogsColumns[42]},
 			},
 			{
 				Name:    "usagelog_group_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[42]},
+				Columns: []*schema.Column{UsageLogsColumns[43]},
 			},
 			{
 				Name:    "usagelog_subscription_id",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[44]},
+				Columns: []*schema.Column{UsageLogsColumns[45]},
 			},
 			{
 				Name:    "usagelog_created_at",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[39]},
+				Columns: []*schema.Column{UsageLogsColumns[40]},
 			},
 			{
 				Name:    "usagelog_model",
@@ -1669,17 +1672,17 @@ var (
 			{
 				Name:    "usagelog_user_id_created_at",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[43], UsageLogsColumns[39]},
+				Columns: []*schema.Column{UsageLogsColumns[44], UsageLogsColumns[40]},
 			},
 			{
 				Name:    "usagelog_api_key_id_created_at",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[40], UsageLogsColumns[39]},
+				Columns: []*schema.Column{UsageLogsColumns[41], UsageLogsColumns[40]},
 			},
 			{
 				Name:    "usagelog_group_id_created_at",
 				Unique:  false,
-				Columns: []*schema.Column{UsageLogsColumns[42], UsageLogsColumns[39]},
+				Columns: []*schema.Column{UsageLogsColumns[43], UsageLogsColumns[40]},
 			},
 		},
 	}

backend/ent/mutation.go

@@ -20842,6 +20842,8 @@ type GroupMutation struct {
 	addvideo_price_720p                     *float64
 	video_price_1080p                       *float64
 	addvideo_price_1080p                    *float64
+	web_search_price_per_call               *float64
+	addweb_search_price_per_call            *float64
 	claude_code_only                        *bool
 	fallback_group_id                       *int64
 	addfallback_group_id                    *int64
@@ -22608,6 +22610,76 @@ func (m *GroupMutation) ResetVideoPrice1080p() {
 	delete(m.clearedFields, group.FieldVideoPrice1080p)
 }
 
+// SetWebSearchPricePerCall sets the "web_search_price_per_call" field.
+func (m *GroupMutation) SetWebSearchPricePerCall(f float64) {
+	m.web_search_price_per_call = &f
+	m.addweb_search_price_per_call = nil
+}
+
+// WebSearchPricePerCall returns the value of the "web_search_price_per_call" field in the mutation.
+func (m *GroupMutation) WebSearchPricePerCall() (r float64, exists bool) {
+	v := m.web_search_price_per_call
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// OldWebSearchPricePerCall returns the old "web_search_price_per_call" 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) OldWebSearchPricePerCall(ctx context.Context) (v *float64, err error) {
+	if !m.op.Is(OpUpdateOne) {
+		return v, errors.New("OldWebSearchPricePerCall is only allowed on UpdateOne operations")
+	}
+	if m.id == nil || m.oldValue == nil {
+		return v, errors.New("OldWebSearchPricePerCall requires an ID field in the mutation")
+	}
+	oldValue, err := m.oldValue(ctx)
+	if err != nil {
+		return v, fmt.Errorf("querying old value for OldWebSearchPricePerCall: %w", err)
+	}
+	return oldValue.WebSearchPricePerCall, nil
+}
+
+// AddWebSearchPricePerCall adds f to the "web_search_price_per_call" field.
+func (m *GroupMutation) AddWebSearchPricePerCall(f float64) {
+	if m.addweb_search_price_per_call != nil {
+		*m.addweb_search_price_per_call += f
+	} else {
+		m.addweb_search_price_per_call = &f
+	}
+}
+
+// AddedWebSearchPricePerCall returns the value that was added to the "web_search_price_per_call" field in this mutation.
+func (m *GroupMutation) AddedWebSearchPricePerCall() (r float64, exists bool) {
+	v := m.addweb_search_price_per_call
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// ClearWebSearchPricePerCall clears the value of the "web_search_price_per_call" field.
+func (m *GroupMutation) ClearWebSearchPricePerCall() {
+	m.web_search_price_per_call = nil
+	m.addweb_search_price_per_call = nil
+	m.clearedFields[group.FieldWebSearchPricePerCall] = struct{}{}
+}
+
+// WebSearchPricePerCallCleared returns if the "web_search_price_per_call" field was cleared in this mutation.
+func (m *GroupMutation) WebSearchPricePerCallCleared() bool {
+	_, ok := m.clearedFields[group.FieldWebSearchPricePerCall]
+	return ok
+}
+
+// ResetWebSearchPricePerCall resets all changes to the "web_search_price_per_call" field.
+func (m *GroupMutation) ResetWebSearchPricePerCall() {
+	m.web_search_price_per_call = nil
+	m.addweb_search_price_per_call = nil
+	delete(m.clearedFields, group.FieldWebSearchPricePerCall)
+}
+
 // SetClaudeCodeOnly sets the "claude_code_only" field.
 func (m *GroupMutation) SetClaudeCodeOnly(b bool) {
 	m.claude_code_only = &b
@@ -23642,7 +23714,7 @@ 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, 47)
+	fields := make([]string, 0, 48)
 	if m.created_at != nil {
 		fields = append(fields, group.FieldCreatedAt)
 	}
@@ -23739,6 +23811,9 @@ func (m *GroupMutation) Fields() []string {
 	if m.video_price_1080p != nil {
 		fields = append(fields, group.FieldVideoPrice1080p)
 	}
+	if m.web_search_price_per_call != nil {
+		fields = append(fields, group.FieldWebSearchPricePerCall)
+	}
 	if m.claude_code_only != nil {
 		fields = append(fields, group.FieldClaudeCodeOnly)
 	}
@@ -23856,6 +23931,8 @@ func (m *GroupMutation) Field(name string) (ent.Value, bool) {
 		return m.VideoPrice720p()
 	case group.FieldVideoPrice1080p:
 		return m.VideoPrice1080p()
+	case group.FieldWebSearchPricePerCall:
+		return m.WebSearchPricePerCall()
 	case group.FieldClaudeCodeOnly:
 		return m.ClaudeCodeOnly()
 	case group.FieldFallbackGroupID:
@@ -23959,6 +24036,8 @@ func (m *GroupMutation) OldField(ctx context.Context, name string) (ent.Value, e
 		return m.OldVideoPrice720p(ctx)
 	case group.FieldVideoPrice1080p:
 		return m.OldVideoPrice1080p(ctx)
+	case group.FieldWebSearchPricePerCall:
+		return m.OldWebSearchPricePerCall(ctx)
 	case group.FieldClaudeCodeOnly:
 		return m.OldClaudeCodeOnly(ctx)
 	case group.FieldFallbackGroupID:
@@ -24222,6 +24301,13 @@ func (m *GroupMutation) SetField(name string, value ent.Value) error {
 		}
 		m.SetVideoPrice1080p(v)
 		return nil
+	case group.FieldWebSearchPricePerCall:
+		v, ok := value.(float64)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.SetWebSearchPricePerCall(v)
+		return nil
 	case group.FieldClaudeCodeOnly:
 		v, ok := value.(bool)
 		if !ok {
@@ -24383,6 +24469,9 @@ func (m *GroupMutation) AddedFields() []string {
 	if m.addvideo_price_1080p != nil {
 		fields = append(fields, group.FieldVideoPrice1080p)
 	}
+	if m.addweb_search_price_per_call != nil {
+		fields = append(fields, group.FieldWebSearchPricePerCall)
+	}
 	if m.addfallback_group_id != nil {
 		fields = append(fields, group.FieldFallbackGroupID)
 	}
@@ -24435,6 +24524,8 @@ func (m *GroupMutation) AddedField(name string) (ent.Value, bool) {
 		return m.AddedVideoPrice720p()
 	case group.FieldVideoPrice1080p:
 		return m.AddedVideoPrice1080p()
+	case group.FieldWebSearchPricePerCall:
+		return m.AddedWebSearchPricePerCall()
 	case group.FieldFallbackGroupID:
 		return m.AddedFallbackGroupID()
 	case group.FieldFallbackGroupIDOnInvalidRequest:
@@ -24564,6 +24655,13 @@ func (m *GroupMutation) AddField(name string, value ent.Value) error {
 		}
 		m.AddVideoPrice1080p(v)
 		return nil
+	case group.FieldWebSearchPricePerCall:
+		v, ok := value.(float64)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.AddWebSearchPricePerCall(v)
+		return nil
 	case group.FieldFallbackGroupID:
 		v, ok := value.(int64)
 		if !ok {
@@ -24633,6 +24731,9 @@ func (m *GroupMutation) ClearedFields() []string {
 	if m.FieldCleared(group.FieldVideoPrice1080p) {
 		fields = append(fields, group.FieldVideoPrice1080p)
 	}
+	if m.FieldCleared(group.FieldWebSearchPricePerCall) {
+		fields = append(fields, group.FieldWebSearchPricePerCall)
+	}
 	if m.FieldCleared(group.FieldFallbackGroupID) {
 		fields = append(fields, group.FieldFallbackGroupID)
 	}
@@ -24689,6 +24790,9 @@ func (m *GroupMutation) ClearField(name string) error {
 	case group.FieldVideoPrice1080p:
 		m.ClearVideoPrice1080p()
 		return nil
+	case group.FieldWebSearchPricePerCall:
+		m.ClearWebSearchPricePerCall()
+		return nil
 	case group.FieldFallbackGroupID:
 		m.ClearFallbackGroupID()
 		return nil
@@ -24802,6 +24906,9 @@ func (m *GroupMutation) ResetField(name string) error {
 	case group.FieldVideoPrice1080p:
 		m.ResetVideoPrice1080p()
 		return nil
+	case group.FieldWebSearchPricePerCall:
+		m.ResetWebSearchPricePerCall()
+		return nil
 	case group.FieldClaudeCodeOnly:
 		m.ResetClaudeCodeOnly()
 		return nil
@@ -38045,6 +38152,7 @@ type SubscriptionPlanMutation struct {
 	addprice          *float64
 	original_price    *float64
 	addoriginal_price *float64
+	currency          *string
 	validity_days     *int
 	addvalidity_days  *int
 	validity_unit     *string
@@ -38413,6 +38521,42 @@ func (m *SubscriptionPlanMutation) ResetOriginalPrice() {
 	delete(m.clearedFields, subscriptionplan.FieldOriginalPrice)
 }
 
+// SetCurrency sets the "currency" field.
+func (m *SubscriptionPlanMutation) SetCurrency(s string) {
+	m.currency = &s
+}
+
+// Currency returns the value of the "currency" field in the mutation.
+func (m *SubscriptionPlanMutation) Currency() (r string, exists bool) {
+	v := m.currency
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// OldCurrency returns the old "currency" field's value of the SubscriptionPlan entity.
+// If the SubscriptionPlan 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 *SubscriptionPlanMutation) OldCurrency(ctx context.Context) (v string, err error) {
+	if !m.op.Is(OpUpdateOne) {
+		return v, errors.New("OldCurrency is only allowed on UpdateOne operations")
+	}
+	if m.id == nil || m.oldValue == nil {
+		return v, errors.New("OldCurrency requires an ID field in the mutation")
+	}
+	oldValue, err := m.oldValue(ctx)
+	if err != nil {
+		return v, fmt.Errorf("querying old value for OldCurrency: %w", err)
+	}
+	return oldValue.Currency, nil
+}
+
+// ResetCurrency resets all changes to the "currency" field.
+func (m *SubscriptionPlanMutation) ResetCurrency() {
+	m.currency = nil
+}
+
 // SetValidityDays sets the "validity_days" field.
 func (m *SubscriptionPlanMutation) SetValidityDays(i int) {
 	m.validity_days = &i
@@ -38775,7 +38919,7 @@ func (m *SubscriptionPlanMutation) Type() string {
 // order to get all numeric fields that were incremented/decremented, call
 // AddedFields().
 func (m *SubscriptionPlanMutation) Fields() []string {
-	fields := make([]string, 0, 13)
+	fields := make([]string, 0, 14)
 	if m.group_id != nil {
 		fields = append(fields, subscriptionplan.FieldGroupID)
 	}
@@ -38791,6 +38935,9 @@ func (m *SubscriptionPlanMutation) Fields() []string {
 	if m.original_price != nil {
 		fields = append(fields, subscriptionplan.FieldOriginalPrice)
 	}
+	if m.currency != nil {
+		fields = append(fields, subscriptionplan.FieldCurrency)
+	}
 	if m.validity_days != nil {
 		fields = append(fields, subscriptionplan.FieldValidityDays)
 	}
@@ -38833,6 +38980,8 @@ func (m *SubscriptionPlanMutation) Field(name string) (ent.Value, bool) {
 		return m.Price()
 	case subscriptionplan.FieldOriginalPrice:
 		return m.OriginalPrice()
+	case subscriptionplan.FieldCurrency:
+		return m.Currency()
 	case subscriptionplan.FieldValidityDays:
 		return m.ValidityDays()
 	case subscriptionplan.FieldValidityUnit:
@@ -38868,6 +39017,8 @@ func (m *SubscriptionPlanMutation) OldField(ctx context.Context, name string) (e
 		return m.OldPrice(ctx)
 	case subscriptionplan.FieldOriginalPrice:
 		return m.OldOriginalPrice(ctx)
+	case subscriptionplan.FieldCurrency:
+		return m.OldCurrency(ctx)
 	case subscriptionplan.FieldValidityDays:
 		return m.OldValidityDays(ctx)
 	case subscriptionplan.FieldValidityUnit:
@@ -38928,6 +39079,13 @@ func (m *SubscriptionPlanMutation) SetField(name string, value ent.Value) error
 		}
 		m.SetOriginalPrice(v)
 		return nil
+	case subscriptionplan.FieldCurrency:
+		v, ok := value.(string)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.SetCurrency(v)
+		return nil
 	case subscriptionplan.FieldValidityDays:
 		v, ok := value.(int)
 		if !ok {
@@ -39120,6 +39278,9 @@ func (m *SubscriptionPlanMutation) ResetField(name string) error {
 	case subscriptionplan.FieldOriginalPrice:
 		m.ResetOriginalPrice()
 		return nil
+	case subscriptionplan.FieldCurrency:
+		m.ResetCurrency()
+		return nil
 	case subscriptionplan.FieldValidityDays:
 		m.ResetValidityDays()
 		return nil
@@ -41656,83 +41817,84 @@ func (m *UsageCleanupTaskMutation) ResetEdge(name string) error {
 // UsageLogMutation represents an operation that mutates the UsageLog nodes in the graph.
 type UsageLogMutation struct {
 	config
-	op                          Op
-	typ                         string
-	id                          *int64
-	request_id                  *string
-	model                       *string
-	requested_model             *string
-	upstream_model              *string
-	channel_id                  *int64
-	addchannel_id               *int64
-	model_mapping_chain         *string
-	billing_tier                *string
-	billing_mode                *string
-	input_tokens                *int
-	addinput_tokens             *int
-	output_tokens               *int
-	addoutput_tokens            *int
-	cache_creation_tokens       *int
-	addcache_creation_tokens    *int
-	cache_read_tokens           *int
-	addcache_read_tokens        *int
-	cache_creation_5m_tokens    *int
-	addcache_creation_5m_tokens *int
-	cache_creation_1h_tokens    *int
-	addcache_creation_1h_tokens *int
-	input_cost                  *float64
-	addinput_cost               *float64
-	output_cost                 *float64
-	addoutput_cost              *float64
-	cache_creation_cost         *float64
-	addcache_creation_cost      *float64
-	cache_read_cost             *float64
-	addcache_read_cost          *float64
-	total_cost                  *float64
-	addtotal_cost               *float64
-	actual_cost                 *float64
-	addactual_cost              *float64
-	rate_multiplier             *float64
-	addrate_multiplier          *float64
-	account_rate_multiplier     *float64
-	addaccount_rate_multiplier  *float64
-	billing_type                *int8
-	addbilling_type             *int8
-	stream                      *bool
-	duration_ms                 *int
-	addduration_ms              *int
-	first_token_ms              *int
-	addfirst_token_ms           *int
-	user_agent                  *string
-	ip_address                  *string
-	image_count                 *int
-	addimage_count              *int
-	image_size                  *string
-	image_input_size            *string
-	image_output_size           *string
-	image_size_source           *string
-	image_size_breakdown        *map[string]int
-	video_count                 *int
-	addvideo_count              *int
-	video_resolution            *string
-	video_duration_seconds      *int
-	addvideo_duration_seconds   *int
-	cache_ttl_overridden        *bool
-	created_at                  *time.Time
-	clearedFields               map[string]struct{}
-	user                        *int64
-	cleareduser                 bool
-	api_key                     *int64
-	clearedapi_key              bool
-	account                     *int64
-	clearedaccount              bool
-	group                       *int64
-	clearedgroup                bool
-	subscription                *int64
-	clearedsubscription         bool
-	done                        bool
-	oldValue                    func(context.Context) (*UsageLog, error)
-	predicates                  []predicate.UsageLog
+	op                           Op
+	typ                          string
+	id                           *int64
+	request_id                   *string
+	model                        *string
+	requested_model              *string
+	upstream_model               *string
+	channel_id                   *int64
+	addchannel_id                *int64
+	model_mapping_chain          *string
+	billing_tier                 *string
+	billing_mode                 *string
+	input_tokens                 *int
+	addinput_tokens              *int
+	output_tokens                *int
+	addoutput_tokens             *int
+	cache_creation_tokens        *int
+	addcache_creation_tokens     *int
+	cache_read_tokens            *int
+	addcache_read_tokens         *int
+	cache_creation_5m_tokens     *int
+	addcache_creation_5m_tokens  *int
+	cache_creation_1h_tokens     *int
+	addcache_creation_1h_tokens  *int
+	input_cost                   *float64
+	addinput_cost                *float64
+	output_cost                  *float64
+	addoutput_cost               *float64
+	cache_creation_cost          *float64
+	addcache_creation_cost       *float64
+	cache_read_cost              *float64
+	addcache_read_cost           *float64
+	total_cost                   *float64
+	addtotal_cost                *float64
+	actual_cost                  *float64
+	addactual_cost               *float64
+	rate_multiplier              *float64
+	addrate_multiplier           *float64
+	long_context_billing_applied *bool
+	account_rate_multiplier      *float64
+	addaccount_rate_multiplier   *float64
+	billing_type                 *int8
+	addbilling_type              *int8
+	stream                       *bool
+	duration_ms                  *int
+	addduration_ms               *int
+	first_token_ms               *int
+	addfirst_token_ms            *int
+	user_agent                   *string
+	ip_address                   *string
+	image_count                  *int
+	addimage_count               *int
+	image_size                   *string
+	image_input_size             *string
+	image_output_size            *string
+	image_size_source            *string
+	image_size_breakdown         *map[string]int
+	video_count                  *int
+	addvideo_count               *int
+	video_resolution             *string
+	video_duration_seconds       *int
+	addvideo_duration_seconds    *int
+	cache_ttl_overridden         *bool
+	created_at                   *time.Time
+	clearedFields                map[string]struct{}
+	user                         *int64
+	cleareduser                  bool
+	api_key                      *int64
+	clearedapi_key               bool
+	account                      *int64
+	clearedaccount               bool
+	group                        *int64
+	clearedgroup                 bool
+	subscription                 *int64
+	clearedsubscription          bool
+	done                         bool
+	oldValue                     func(context.Context) (*UsageLog, error)
+	predicates                   []predicate.UsageLog
 }
 
 var _ ent.Mutation = (*UsageLogMutation)(nil)
@@ -43154,6 +43316,42 @@ func (m *UsageLogMutation) ResetRateMultiplier() {
 	m.addrate_multiplier = nil
 }
 
+// SetLongContextBillingApplied sets the "long_context_billing_applied" field.
+func (m *UsageLogMutation) SetLongContextBillingApplied(b bool) {
+	m.long_context_billing_applied = &b
+}
+
+// LongContextBillingApplied returns the value of the "long_context_billing_applied" field in the mutation.
+func (m *UsageLogMutation) LongContextBillingApplied() (r bool, exists bool) {
+	v := m.long_context_billing_applied
+	if v == nil {
+		return
+	}
+	return *v, true
+}
+
+// OldLongContextBillingApplied returns the old "long_context_billing_applied" field's value of the UsageLog entity.
+// If the UsageLog 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 *UsageLogMutation) OldLongContextBillingApplied(ctx context.Context) (v bool, err error) {
+	if !m.op.Is(OpUpdateOne) {
+		return v, errors.New("OldLongContextBillingApplied is only allowed on UpdateOne operations")
+	}
+	if m.id == nil || m.oldValue == nil {
+		return v, errors.New("OldLongContextBillingApplied requires an ID field in the mutation")
+	}
+	oldValue, err := m.oldValue(ctx)
+	if err != nil {
+		return v, fmt.Errorf("querying old value for OldLongContextBillingApplied: %w", err)
+	}
+	return oldValue.LongContextBillingApplied, nil
+}
+
+// ResetLongContextBillingApplied resets all changes to the "long_context_billing_applied" field.
+func (m *UsageLogMutation) ResetLongContextBillingApplied() {
+	m.long_context_billing_applied = nil
+}
+
 // SetAccountRateMultiplier sets the "account_rate_multiplier" field.
 func (m *UsageLogMutation) SetAccountRateMultiplier(f float64) {
 	m.account_rate_multiplier = &f
@@ -44271,7 +44469,7 @@ func (m *UsageLogMutation) Type() string {
 // order to get all numeric fields that were incremented/decremented, call
 // AddedFields().
 func (m *UsageLogMutation) Fields() []string {
-	fields := make([]string, 0, 44)
+	fields := make([]string, 0, 45)
 	if m.user != nil {
 		fields = append(fields, usagelog.FieldUserID)
 	}
@@ -44350,6 +44548,9 @@ func (m *UsageLogMutation) Fields() []string {
 	if m.rate_multiplier != nil {
 		fields = append(fields, usagelog.FieldRateMultiplier)
 	}
+	if m.long_context_billing_applied != nil {
+		fields = append(fields, usagelog.FieldLongContextBillingApplied)
+	}
 	if m.account_rate_multiplier != nil {
 		fields = append(fields, usagelog.FieldAccountRateMultiplier)
 	}
@@ -44464,6 +44665,8 @@ func (m *UsageLogMutation) Field(name string) (ent.Value, bool) {
 		return m.ActualCost()
 	case usagelog.FieldRateMultiplier:
 		return m.RateMultiplier()
+	case usagelog.FieldLongContextBillingApplied:
+		return m.LongContextBillingApplied()
 	case usagelog.FieldAccountRateMultiplier:
 		return m.AccountRateMultiplier()
 	case usagelog.FieldBillingType:
@@ -44561,6 +44764,8 @@ func (m *UsageLogMutation) OldField(ctx context.Context, name string) (ent.Value
 		return m.OldActualCost(ctx)
 	case usagelog.FieldRateMultiplier:
 		return m.OldRateMultiplier(ctx)
+	case usagelog.FieldLongContextBillingApplied:
+		return m.OldLongContextBillingApplied(ctx)
 	case usagelog.FieldAccountRateMultiplier:
 		return m.OldAccountRateMultiplier(ctx)
 	case usagelog.FieldBillingType:
@@ -44788,6 +44993,13 @@ func (m *UsageLogMutation) SetField(name string, value ent.Value) error {
 		}
 		m.SetRateMultiplier(v)
 		return nil
+	case usagelog.FieldLongContextBillingApplied:
+		v, ok := value.(bool)
+		if !ok {
+			return fmt.Errorf("unexpected type %T for field %s", value, name)
+		}
+		m.SetLongContextBillingApplied(v)
+		return nil
 	case usagelog.FieldAccountRateMultiplier:
 		v, ok := value.(float64)
 		if !ok {
@@ -45419,6 +45631,9 @@ func (m *UsageLogMutation) ResetField(name string) error {
 	case usagelog.FieldRateMultiplier:
 		m.ResetRateMultiplier()
 		return nil
+	case usagelog.FieldLongContextBillingApplied:
+		m.ResetLongContextBillingApplied()
+		return nil
 	case usagelog.FieldAccountRateMultiplier:
 		m.ResetAccountRateMultiplier()
 		return nil

backend/ent/runtime/runtime.go

@@ -1044,53 +1044,53 @@ func init() {
 	// group.DefaultVideoRateMultiplier holds the default value on creation for the video_rate_multiplier field.
 	group.DefaultVideoRateMultiplier = groupDescVideoRateMultiplier.Default.(float64)
 	// groupDescClaudeCodeOnly is the schema descriptor for claude_code_only field.
-	groupDescClaudeCodeOnly := groupFields[29].Descriptor()
+	groupDescClaudeCodeOnly := groupFields[30].Descriptor()
 	// group.DefaultClaudeCodeOnly holds the default value on creation for the claude_code_only field.
 	group.DefaultClaudeCodeOnly = groupDescClaudeCodeOnly.Default.(bool)
 	// groupDescModelRoutingEnabled is the schema descriptor for model_routing_enabled field.
-	groupDescModelRoutingEnabled := groupFields[33].Descriptor()
+	groupDescModelRoutingEnabled := groupFields[34].Descriptor()
 	// group.DefaultModelRoutingEnabled holds the default value on creation for the model_routing_enabled field.
 	group.DefaultModelRoutingEnabled = groupDescModelRoutingEnabled.Default.(bool)
 	// groupDescMcpXMLInject is the schema descriptor for mcp_xml_inject field.
-	groupDescMcpXMLInject := groupFields[34].Descriptor()
+	groupDescMcpXMLInject := groupFields[35].Descriptor()
 	// group.DefaultMcpXMLInject holds the default value on creation for the mcp_xml_inject field.
 	group.DefaultMcpXMLInject = groupDescMcpXMLInject.Default.(bool)
 	// groupDescSupportedModelScopes is the schema descriptor for supported_model_scopes field.
-	groupDescSupportedModelScopes := groupFields[35].Descriptor()
+	groupDescSupportedModelScopes := groupFields[36].Descriptor()
 	// group.DefaultSupportedModelScopes holds the default value on creation for the supported_model_scopes field.
 	group.DefaultSupportedModelScopes = groupDescSupportedModelScopes.Default.([]string)
 	// groupDescSortOrder is the schema descriptor for sort_order field.
-	groupDescSortOrder := groupFields[36].Descriptor()
+	groupDescSortOrder := groupFields[37].Descriptor()
 	// group.DefaultSortOrder holds the default value on creation for the sort_order field.
 	group.DefaultSortOrder = groupDescSortOrder.Default.(int)
 	// groupDescAllowMessagesDispatch is the schema descriptor for allow_messages_dispatch field.
-	groupDescAllowMessagesDispatch := groupFields[37].Descriptor()
+	groupDescAllowMessagesDispatch := groupFields[38].Descriptor()
 	// group.DefaultAllowMessagesDispatch holds the default value on creation for the allow_messages_dispatch field.
 	group.DefaultAllowMessagesDispatch = groupDescAllowMessagesDispatch.Default.(bool)
 	// groupDescRequireOauthOnly is the schema descriptor for require_oauth_only field.
-	groupDescRequireOauthOnly := groupFields[38].Descriptor()
+	groupDescRequireOauthOnly := groupFields[39].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[39].Descriptor()
+	groupDescRequirePrivacySet := groupFields[40].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[40].Descriptor()
+	groupDescDefaultMappedModel := groupFields[41].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[41].Descriptor()
+	groupDescMessagesDispatchModelConfig := groupFields[42].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[42].Descriptor()
+	groupDescModelsListConfig := groupFields[43].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[43].Descriptor()
+	groupDescRpmLimit := groupFields[44].Descriptor()
 	// group.DefaultRpmLimit holds the default value on creation for the rpm_limit field.
 	group.DefaultRpmLimit = groupDescRpmLimit.Default.(int)
 	idempotencyrecordMixin := schema.IdempotencyRecord{}.Mixin()
@@ -1718,40 +1718,46 @@ func init() {
 	subscriptionplanDescDescription := subscriptionplanFields[2].Descriptor()
 	// subscriptionplan.DefaultDescription holds the default value on creation for the description field.
 	subscriptionplan.DefaultDescription = subscriptionplanDescDescription.Default.(string)
+	// subscriptionplanDescCurrency is the schema descriptor for currency field.
+	subscriptionplanDescCurrency := subscriptionplanFields[5].Descriptor()
+	// subscriptionplan.DefaultCurrency holds the default value on creation for the currency field.
+	subscriptionplan.DefaultCurrency = subscriptionplanDescCurrency.Default.(string)
+	// subscriptionplan.CurrencyValidator is a validator for the "currency" field. It is called by the builders before save.
+	subscriptionplan.CurrencyValidator = subscriptionplanDescCurrency.Validators[0].(func(string) error)
 	// subscriptionplanDescValidityDays is the schema descriptor for validity_days field.
-	subscriptionplanDescValidityDays := subscriptionplanFields[5].Descriptor()
+	subscriptionplanDescValidityDays := subscriptionplanFields[6].Descriptor()
 	// subscriptionplan.DefaultValidityDays holds the default value on creation for the validity_days field.
 	subscriptionplan.DefaultValidityDays = subscriptionplanDescValidityDays.Default.(int)
 	// subscriptionplanDescValidityUnit is the schema descriptor for validity_unit field.
-	subscriptionplanDescValidityUnit := subscriptionplanFields[6].Descriptor()
+	subscriptionplanDescValidityUnit := subscriptionplanFields[7].Descriptor()
 	// subscriptionplan.DefaultValidityUnit holds the default value on creation for the validity_unit field.
 	subscriptionplan.DefaultValidityUnit = subscriptionplanDescValidityUnit.Default.(string)
 	// subscriptionplan.ValidityUnitValidator is a validator for the "validity_unit" field. It is called by the builders before save.
 	subscriptionplan.ValidityUnitValidator = subscriptionplanDescValidityUnit.Validators[0].(func(string) error)
 	// subscriptionplanDescFeatures is the schema descriptor for features field.
-	subscriptionplanDescFeatures := subscriptionplanFields[7].Descriptor()
+	subscriptionplanDescFeatures := subscriptionplanFields[8].Descriptor()
 	// subscriptionplan.DefaultFeatures holds the default value on creation for the features field.
 	subscriptionplan.DefaultFeatures = subscriptionplanDescFeatures.Default.(string)
 	// subscriptionplanDescProductName is the schema descriptor for product_name field.
-	subscriptionplanDescProductName := subscriptionplanFields[8].Descriptor()
+	subscriptionplanDescProductName := subscriptionplanFields[9].Descriptor()
 	// subscriptionplan.DefaultProductName holds the default value on creation for the product_name field.
 	subscriptionplan.DefaultProductName = subscriptionplanDescProductName.Default.(string)
 	// subscriptionplan.ProductNameValidator is a validator for the "product_name" field. It is called by the builders before save.
 	subscriptionplan.ProductNameValidator = subscriptionplanDescProductName.Validators[0].(func(string) error)
 	// subscriptionplanDescForSale is the schema descriptor for for_sale field.
-	subscriptionplanDescForSale := subscriptionplanFields[9].Descriptor()
+	subscriptionplanDescForSale := subscriptionplanFields[10].Descriptor()
 	// subscriptionplan.DefaultForSale holds the default value on creation for the for_sale field.
 	subscriptionplan.DefaultForSale = subscriptionplanDescForSale.Default.(bool)
 	// subscriptionplanDescSortOrder is the schema descriptor for sort_order field.
-	subscriptionplanDescSortOrder := subscriptionplanFields[10].Descriptor()
+	subscriptionplanDescSortOrder := subscriptionplanFields[11].Descriptor()
 	// subscriptionplan.DefaultSortOrder holds the default value on creation for the sort_order field.
 	subscriptionplan.DefaultSortOrder = subscriptionplanDescSortOrder.Default.(int)
 	// subscriptionplanDescCreatedAt is the schema descriptor for created_at field.
-	subscriptionplanDescCreatedAt := subscriptionplanFields[11].Descriptor()
+	subscriptionplanDescCreatedAt := subscriptionplanFields[12].Descriptor()
 	// subscriptionplan.DefaultCreatedAt holds the default value on creation for the created_at field.
 	subscriptionplan.DefaultCreatedAt = subscriptionplanDescCreatedAt.Default.(func() time.Time)
 	// subscriptionplanDescUpdatedAt is the schema descriptor for updated_at field.
-	subscriptionplanDescUpdatedAt := subscriptionplanFields[12].Descriptor()
+	subscriptionplanDescUpdatedAt := subscriptionplanFields[13].Descriptor()
 	// subscriptionplan.DefaultUpdatedAt holds the default value on creation for the updated_at field.
 	subscriptionplan.DefaultUpdatedAt = subscriptionplanDescUpdatedAt.Default.(func() time.Time)
 	// subscriptionplan.UpdateDefaultUpdatedAt holds the default value on update for the updated_at field.
@@ -1940,56 +1946,60 @@ func init() {
 	usagelogDescRateMultiplier := usagelogFields[25].Descriptor()
 	// usagelog.DefaultRateMultiplier holds the default value on creation for the rate_multiplier field.
 	usagelog.DefaultRateMultiplier = usagelogDescRateMultiplier.Default.(float64)
+	// usagelogDescLongContextBillingApplied is the schema descriptor for long_context_billing_applied field.
+	usagelogDescLongContextBillingApplied := usagelogFields[26].Descriptor()
+	// usagelog.DefaultLongContextBillingApplied holds the default value on creation for the long_context_billing_applied field.
+	usagelog.DefaultLongContextBillingApplied = usagelogDescLongContextBillingApplied.Default.(bool)
 	// usagelogDescBillingType is the schema descriptor for billing_type field.
-	usagelogDescBillingType := usagelogFields[27].Descriptor()
+	usagelogDescBillingType := usagelogFields[28].Descriptor()
 	// usagelog.DefaultBillingType holds the default value on creation for the billing_type field.
 	usagelog.DefaultBillingType = usagelogDescBillingType.Default.(int8)
 	// usagelogDescStream is the schema descriptor for stream field.
-	usagelogDescStream := usagelogFields[28].Descriptor()
+	usagelogDescStream := usagelogFields[29].Descriptor()
 	// usagelog.DefaultStream holds the default value on creation for the stream field.
 	usagelog.DefaultStream = usagelogDescStream.Default.(bool)
 	// usagelogDescUserAgent is the schema descriptor for user_agent field.
-	usagelogDescUserAgent := usagelogFields[31].Descriptor()
+	usagelogDescUserAgent := usagelogFields[32].Descriptor()
 	// usagelog.UserAgentValidator is a validator for the "user_agent" field. It is called by the builders before save.
 	usagelog.UserAgentValidator = usagelogDescUserAgent.Validators[0].(func(string) error)
 	// usagelogDescIPAddress is the schema descriptor for ip_address field.
-	usagelogDescIPAddress := usagelogFields[32].Descriptor()
+	usagelogDescIPAddress := usagelogFields[33].Descriptor()
 	// usagelog.IPAddressValidator is a validator for the "ip_address" field. It is called by the builders before save.
 	usagelog.IPAddressValidator = usagelogDescIPAddress.Validators[0].(func(string) error)
 	// usagelogDescImageCount is the schema descriptor for image_count field.
-	usagelogDescImageCount := usagelogFields[33].Descriptor()
+	usagelogDescImageCount := usagelogFields[34].Descriptor()
 	// usagelog.DefaultImageCount holds the default value on creation for the image_count field.
 	usagelog.DefaultImageCount = usagelogDescImageCount.Default.(int)
 	// usagelogDescImageSize is the schema descriptor for image_size field.
-	usagelogDescImageSize := usagelogFields[34].Descriptor()
+	usagelogDescImageSize := usagelogFields[35].Descriptor()
 	// usagelog.ImageSizeValidator is a validator for the "image_size" field. It is called by the builders before save.
 	usagelog.ImageSizeValidator = usagelogDescImageSize.Validators[0].(func(string) error)
 	// usagelogDescImageInputSize is the schema descriptor for image_input_size field.
-	usagelogDescImageInputSize := usagelogFields[35].Descriptor()
+	usagelogDescImageInputSize := usagelogFields[36].Descriptor()
 	// usagelog.ImageInputSizeValidator is a validator for the "image_input_size" field. It is called by the builders before save.
 	usagelog.ImageInputSizeValidator = usagelogDescImageInputSize.Validators[0].(func(string) error)
 	// usagelogDescImageOutputSize is the schema descriptor for image_output_size field.
-	usagelogDescImageOutputSize := usagelogFields[36].Descriptor()
+	usagelogDescImageOutputSize := usagelogFields[37].Descriptor()
 	// usagelog.ImageOutputSizeValidator is a validator for the "image_output_size" field. It is called by the builders before save.
 	usagelog.ImageOutputSizeValidator = usagelogDescImageOutputSize.Validators[0].(func(string) error)
 	// usagelogDescImageSizeSource is the schema descriptor for image_size_source field.
-	usagelogDescImageSizeSource := usagelogFields[37].Descriptor()
+	usagelogDescImageSizeSource := usagelogFields[38].Descriptor()
 	// usagelog.ImageSizeSourceValidator is a validator for the "image_size_source" field. It is called by the builders before save.
 	usagelog.ImageSizeSourceValidator = usagelogDescImageSizeSource.Validators[0].(func(string) error)
 	// usagelogDescVideoCount is the schema descriptor for video_count field.
-	usagelogDescVideoCount := usagelogFields[39].Descriptor()
+	usagelogDescVideoCount := usagelogFields[40].Descriptor()
 	// usagelog.DefaultVideoCount holds the default value on creation for the video_count field.
 	usagelog.DefaultVideoCount = usagelogDescVideoCount.Default.(int)
 	// usagelogDescVideoResolution is the schema descriptor for video_resolution field.
-	usagelogDescVideoResolution := usagelogFields[40].Descriptor()
+	usagelogDescVideoResolution := usagelogFields[41].Descriptor()
 	// usagelog.VideoResolutionValidator is a validator for the "video_resolution" field. It is called by the builders before save.
 	usagelog.VideoResolutionValidator = usagelogDescVideoResolution.Validators[0].(func(string) error)
 	// usagelogDescCacheTTLOverridden is the schema descriptor for cache_ttl_overridden field.
-	usagelogDescCacheTTLOverridden := usagelogFields[42].Descriptor()
+	usagelogDescCacheTTLOverridden := usagelogFields[43].Descriptor()
 	// usagelog.DefaultCacheTTLOverridden holds the default value on creation for the cache_ttl_overridden field.
 	usagelog.DefaultCacheTTLOverridden = usagelogDescCacheTTLOverridden.Default.(bool)
 	// usagelogDescCreatedAt is the schema descriptor for created_at field.
-	usagelogDescCreatedAt := usagelogFields[43].Descriptor()
+	usagelogDescCreatedAt := usagelogFields[44].Descriptor()
 	// usagelog.DefaultCreatedAt holds the default value on creation for the created_at field.
 	usagelog.DefaultCreatedAt = usagelogDescCreatedAt.Default.(func() time.Time)
 	userMixin := schema.User{}.Mixin()

backend/ent/schema/channel_monitor.go

@@ -35,7 +35,7 @@ func (ChannelMonitor) Fields() []ent.Field {
 			NotEmpty().
 			MaxLen(100),
 		field.Enum("provider").
-			Values("openai", "anthropic", "gemini"),
+			Values("openai", "anthropic", "gemini", "grok"),
 		field.String("api_mode").
 			Default("chat_completions").
 			MaxLen(32).

backend/ent/schema/channel_monitor_request_template.go

@@ -39,7 +39,7 @@ func (ChannelMonitorRequestTemplate) Fields() []ent.Field {
 			NotEmpty().
 			MaxLen(100),
 		field.Enum("provider").
-			Values("openai", "anthropic", "gemini"),
+			Values("openai", "anthropic", "gemini", "grok"),
 		field.String("api_mode").
 			Default("chat_completions").
 			MaxLen(32).

backend/ent/schema/group.go

@@ -142,6 +142,11 @@ func (Group) Fields() []ent.Field {
 			Optional().
 			Nillable().
 			SchemaType(map[string]string{dialect.Postgres: "decimal(20,8)"}),
+		field.Float("web_search_price_per_call").
+			Optional().
+			Nillable().
+			SchemaType(map[string]string{dialect.Postgres: "decimal(20,8)"}).
+			Comment("Codex alpha/search 网页搜索单次价格（USD/次）；nil 表示使用默认价 0.01（官方 $10/1000 次）"),
 
 		// Claude Code 客户端限制 (added by migration 029)
 		field.Bool("claude_code_only").

backend/ent/schema/subscription_plan.go

@@ -43,6 +43,9 @@ func (SubscriptionPlan) Fields() []ent.Field {
 			SchemaType(map[string]string{dialect.Postgres: "decimal(20,2)"}).
 			Optional().
 			Nillable(),
+		field.String("currency").
+			MaxLen(3).
+			Default(""),
 		field.Int("validity_days").
 			Default(30),
 		field.String("validity_unit").

backend/ent/schema/usage_log.go

@@ -100,6 +100,9 @@ func (UsageLog) Fields() []ent.Field {
 		field.Float("rate_multiplier").
 			Default(1).
 			SchemaType(map[string]string{dialect.Postgres: "decimal(10,4)"}),
+		field.Bool("long_context_billing_applied").
+			Default(false).
+			Comment("Whether long-context pricing changed token prices for this request"),
 
 		// account_rate_multiplier: 账号计费倍率快照（NULL 表示按 1.0 处理）
 		field.Float("account_rate_multiplier").

backend/ent/subscriptionplan.go

@@ -27,6 +27,8 @@ type SubscriptionPlan struct {
 	Price float64 `json:"price,omitempty"`
 	// OriginalPrice holds the value of the "original_price" field.
 	OriginalPrice *float64 `json:"original_price,omitempty"`
+	// Currency holds the value of the "currency" field.
+	Currency string `json:"currency,omitempty"`
 	// ValidityDays holds the value of the "validity_days" field.
 	ValidityDays int `json:"validity_days,omitempty"`
 	// ValidityUnit holds the value of the "validity_unit" field.
@@ -57,7 +59,7 @@ func (*SubscriptionPlan) scanValues(columns []string) ([]any, error) {
 			values[i] = new(sql.NullFloat64)
 		case subscriptionplan.FieldID, subscriptionplan.FieldGroupID, subscriptionplan.FieldValidityDays, subscriptionplan.FieldSortOrder:
 			values[i] = new(sql.NullInt64)
-		case subscriptionplan.FieldName, subscriptionplan.FieldDescription, subscriptionplan.FieldValidityUnit, subscriptionplan.FieldFeatures, subscriptionplan.FieldProductName:
+		case subscriptionplan.FieldName, subscriptionplan.FieldDescription, subscriptionplan.FieldCurrency, subscriptionplan.FieldValidityUnit, subscriptionplan.FieldFeatures, subscriptionplan.FieldProductName:
 			values[i] = new(sql.NullString)
 		case subscriptionplan.FieldCreatedAt, subscriptionplan.FieldUpdatedAt:
 			values[i] = new(sql.NullTime)
@@ -113,6 +115,12 @@ func (_m *SubscriptionPlan) assignValues(columns []string, values []any) error {
 				_m.OriginalPrice = new(float64)
 				*_m.OriginalPrice = value.Float64
 			}
+		case subscriptionplan.FieldCurrency:
+			if value, ok := values[i].(*sql.NullString); !ok {
+				return fmt.Errorf("unexpected type %T for field currency", values[i])
+			} else if value.Valid {
+				_m.Currency = value.String
+			}
 		case subscriptionplan.FieldValidityDays:
 			if value, ok := values[i].(*sql.NullInt64); !ok {
 				return fmt.Errorf("unexpected type %T for field validity_days", values[i])
@@ -214,6 +222,9 @@ func (_m *SubscriptionPlan) String() string {
 		builder.WriteString(fmt.Sprintf("%v", *v))
 	}
 	builder.WriteString(", ")
+	builder.WriteString("currency=")
+	builder.WriteString(_m.Currency)
+	builder.WriteString(", ")
 	builder.WriteString("validity_days=")
 	builder.WriteString(fmt.Sprintf("%v", _m.ValidityDays))
 	builder.WriteString(", ")

backend/ent/subscriptionplan/subscriptionplan.go

@@ -23,6 +23,8 @@ const (
 	FieldPrice = "price"
 	// FieldOriginalPrice holds the string denoting the original_price field in the database.
 	FieldOriginalPrice = "original_price"
+	// FieldCurrency holds the string denoting the currency field in the database.
+	FieldCurrency = "currency"
 	// FieldValidityDays holds the string denoting the validity_days field in the database.
 	FieldValidityDays = "validity_days"
 	// FieldValidityUnit holds the string denoting the validity_unit field in the database.
@@ -51,6 +53,7 @@ var Columns = []string{
 	FieldDescription,
 	FieldPrice,
 	FieldOriginalPrice,
+	FieldCurrency,
 	FieldValidityDays,
 	FieldValidityUnit,
 	FieldFeatures,
@@ -76,6 +79,10 @@ var (
 	NameValidator func(string) error
 	// DefaultDescription holds the default value on creation for the "description" field.
 	DefaultDescription string
+	// DefaultCurrency holds the default value on creation for the "currency" field.
+	DefaultCurrency string
+	// CurrencyValidator is a validator for the "currency" field. It is called by the builders before save.
+	CurrencyValidator func(string) error
 	// DefaultValidityDays holds the default value on creation for the "validity_days" field.
 	DefaultValidityDays int
 	// DefaultValidityUnit holds the default value on creation for the "validity_unit" field.
@@ -133,6 +140,11 @@ func ByOriginalPrice(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldOriginalPrice, opts...).ToFunc()
 }
 
+// ByCurrency orders the results by the currency field.
+func ByCurrency(opts ...sql.OrderTermOption) OrderOption {
+	return sql.OrderByField(FieldCurrency, opts...).ToFunc()
+}
+
 // ByValidityDays orders the results by the validity_days field.
 func ByValidityDays(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldValidityDays, opts...).ToFunc()

backend/ent/subscriptionplan/where.go

@@ -79,6 +79,11 @@ func OriginalPrice(v float64) predicate.SubscriptionPlan {
 	return predicate.SubscriptionPlan(sql.FieldEQ(FieldOriginalPrice, v))
 }
 
+// Currency applies equality check predicate on the "currency" field. It's identical to CurrencyEQ.
+func Currency(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldEQ(FieldCurrency, v))
+}
+
 // ValidityDays applies equality check predicate on the "validity_days" field. It's identical to ValidityDaysEQ.
 func ValidityDays(v int) predicate.SubscriptionPlan {
 	return predicate.SubscriptionPlan(sql.FieldEQ(FieldValidityDays, v))
@@ -379,6 +384,71 @@ func OriginalPriceNotNil() predicate.SubscriptionPlan {
 	return predicate.SubscriptionPlan(sql.FieldNotNull(FieldOriginalPrice))
 }
 
+// CurrencyEQ applies the EQ predicate on the "currency" field.
+func CurrencyEQ(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldEQ(FieldCurrency, v))
+}
+
+// CurrencyNEQ applies the NEQ predicate on the "currency" field.
+func CurrencyNEQ(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldNEQ(FieldCurrency, v))
+}
+
+// CurrencyIn applies the In predicate on the "currency" field.
+func CurrencyIn(vs ...string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldIn(FieldCurrency, vs...))
+}
+
+// CurrencyNotIn applies the NotIn predicate on the "currency" field.
+func CurrencyNotIn(vs ...string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldNotIn(FieldCurrency, vs...))
+}
+
+// CurrencyGT applies the GT predicate on the "currency" field.
+func CurrencyGT(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldGT(FieldCurrency, v))
+}
+
+// CurrencyGTE applies the GTE predicate on the "currency" field.
+func CurrencyGTE(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldGTE(FieldCurrency, v))
+}
+
+// CurrencyLT applies the LT predicate on the "currency" field.
+func CurrencyLT(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldLT(FieldCurrency, v))
+}
+
+// CurrencyLTE applies the LTE predicate on the "currency" field.
+func CurrencyLTE(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldLTE(FieldCurrency, v))
+}
+
+// CurrencyContains applies the Contains predicate on the "currency" field.
+func CurrencyContains(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldContains(FieldCurrency, v))
+}
+
+// CurrencyHasPrefix applies the HasPrefix predicate on the "currency" field.
+func CurrencyHasPrefix(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldHasPrefix(FieldCurrency, v))
+}
+
+// CurrencyHasSuffix applies the HasSuffix predicate on the "currency" field.
+func CurrencyHasSuffix(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldHasSuffix(FieldCurrency, v))
+}
+
+// CurrencyEqualFold applies the EqualFold predicate on the "currency" field.
+func CurrencyEqualFold(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldEqualFold(FieldCurrency, v))
+}
+
+// CurrencyContainsFold applies the ContainsFold predicate on the "currency" field.
+func CurrencyContainsFold(v string) predicate.SubscriptionPlan {
+	return predicate.SubscriptionPlan(sql.FieldContainsFold(FieldCurrency, v))
+}
+
 // ValidityDaysEQ applies the EQ predicate on the "validity_days" field.
 func ValidityDaysEQ(v int) predicate.SubscriptionPlan {
 	return predicate.SubscriptionPlan(sql.FieldEQ(FieldValidityDays, v))

backend/ent/subscriptionplan_create.go

@@ -68,6 +68,20 @@ func (_c *SubscriptionPlanCreate) SetNillableOriginalPrice(v *float64) *Subscrip
 	return _c
 }
 
+// SetCurrency sets the "currency" field.
+func (_c *SubscriptionPlanCreate) SetCurrency(v string) *SubscriptionPlanCreate {
+	_c.mutation.SetCurrency(v)
+	return _c
+}
+
+// SetNillableCurrency sets the "currency" field if the given value is not nil.
+func (_c *SubscriptionPlanCreate) SetNillableCurrency(v *string) *SubscriptionPlanCreate {
+	if v != nil {
+		_c.SetCurrency(*v)
+	}
+	return _c
+}
+
 // SetValidityDays sets the "validity_days" field.
 func (_c *SubscriptionPlanCreate) SetValidityDays(v int) *SubscriptionPlanCreate {
 	_c.mutation.SetValidityDays(v)
@@ -219,6 +233,10 @@ func (_c *SubscriptionPlanCreate) defaults() {
 		v := subscriptionplan.DefaultDescription
 		_c.mutation.SetDescription(v)
 	}
+	if _, ok := _c.mutation.Currency(); !ok {
+		v := subscriptionplan.DefaultCurrency
+		_c.mutation.SetCurrency(v)
+	}
 	if _, ok := _c.mutation.ValidityDays(); !ok {
 		v := subscriptionplan.DefaultValidityDays
 		_c.mutation.SetValidityDays(v)
@@ -272,6 +290,14 @@ func (_c *SubscriptionPlanCreate) check() error {
 	if _, ok := _c.mutation.Price(); !ok {
 		return &ValidationError{Name: "price", err: errors.New(`ent: missing required field "SubscriptionPlan.price"`)}
 	}
+	if _, ok := _c.mutation.Currency(); !ok {
+		return &ValidationError{Name: "currency", err: errors.New(`ent: missing required field "SubscriptionPlan.currency"`)}
+	}
+	if v, ok := _c.mutation.Currency(); ok {
+		if err := subscriptionplan.CurrencyValidator(v); err != nil {
+			return &ValidationError{Name: "currency", err: fmt.Errorf(`ent: validator failed for field "SubscriptionPlan.currency": %w`, err)}
+		}
+	}
 	if _, ok := _c.mutation.ValidityDays(); !ok {
 		return &ValidationError{Name: "validity_days", err: errors.New(`ent: missing required field "SubscriptionPlan.validity_days"`)}
 	}
@@ -353,6 +379,10 @@ func (_c *SubscriptionPlanCreate) createSpec() (*SubscriptionPlan, *sqlgraph.Cre
 		_spec.SetField(subscriptionplan.FieldOriginalPrice, field.TypeFloat64, value)
 		_node.OriginalPrice = &value
 	}
+	if value, ok := _c.mutation.Currency(); ok {
+		_spec.SetField(subscriptionplan.FieldCurrency, field.TypeString, value)
+		_node.Currency = value
+	}
 	if value, ok := _c.mutation.ValidityDays(); ok {
 		_spec.SetField(subscriptionplan.FieldValidityDays, field.TypeInt, value)
 		_node.ValidityDays = value
@@ -521,6 +551,18 @@ func (u *SubscriptionPlanUpsert) ClearOriginalPrice() *SubscriptionPlanUpsert {
 	return u
 }
 
+// SetCurrency sets the "currency" field.
+func (u *SubscriptionPlanUpsert) SetCurrency(v string) *SubscriptionPlanUpsert {
+	u.Set(subscriptionplan.FieldCurrency, v)
+	return u
+}
+
+// UpdateCurrency sets the "currency" field to the value that was provided on create.
+func (u *SubscriptionPlanUpsert) UpdateCurrency() *SubscriptionPlanUpsert {
+	u.SetExcluded(subscriptionplan.FieldCurrency)
+	return u
+}
+
 // SetValidityDays sets the "validity_days" field.
 func (u *SubscriptionPlanUpsert) SetValidityDays(v int) *SubscriptionPlanUpsert {
 	u.Set(subscriptionplan.FieldValidityDays, v)
@@ -760,6 +802,20 @@ func (u *SubscriptionPlanUpsertOne) ClearOriginalPrice() *SubscriptionPlanUpsert
 	})
 }
 
+// SetCurrency sets the "currency" field.
+func (u *SubscriptionPlanUpsertOne) SetCurrency(v string) *SubscriptionPlanUpsertOne {
+	return u.Update(func(s *SubscriptionPlanUpsert) {
+		s.SetCurrency(v)
+	})
+}
+
+// UpdateCurrency sets the "currency" field to the value that was provided on create.
+func (u *SubscriptionPlanUpsertOne) UpdateCurrency() *SubscriptionPlanUpsertOne {
+	return u.Update(func(s *SubscriptionPlanUpsert) {
+		s.UpdateCurrency()
+	})
+}
+
 // SetValidityDays sets the "validity_days" field.
 func (u *SubscriptionPlanUpsertOne) SetValidityDays(v int) *SubscriptionPlanUpsertOne {
 	return u.Update(func(s *SubscriptionPlanUpsert) {
@@ -1181,6 +1237,20 @@ func (u *SubscriptionPlanUpsertBulk) ClearOriginalPrice() *SubscriptionPlanUpser
 	})
 }
 
+// SetCurrency sets the "currency" field.
+func (u *SubscriptionPlanUpsertBulk) SetCurrency(v string) *SubscriptionPlanUpsertBulk {
+	return u.Update(func(s *SubscriptionPlanUpsert) {
+		s.SetCurrency(v)
+	})
+}
+
+// UpdateCurrency sets the "currency" field to the value that was provided on create.
+func (u *SubscriptionPlanUpsertBulk) UpdateCurrency() *SubscriptionPlanUpsertBulk {
+	return u.Update(func(s *SubscriptionPlanUpsert) {
+		s.UpdateCurrency()
+	})
+}
+
 // SetValidityDays sets the "validity_days" field.
 func (u *SubscriptionPlanUpsertBulk) SetValidityDays(v int) *SubscriptionPlanUpsertBulk {
 	return u.Update(func(s *SubscriptionPlanUpsert) {

backend/ent/subscriptionplan_update.go

@@ -125,6 +125,20 @@ func (_u *SubscriptionPlanUpdate) ClearOriginalPrice() *SubscriptionPlanUpdate {
 	return _u
 }
 
+// SetCurrency sets the "currency" field.
+func (_u *SubscriptionPlanUpdate) SetCurrency(v string) *SubscriptionPlanUpdate {
+	_u.mutation.SetCurrency(v)
+	return _u
+}
+
+// SetNillableCurrency sets the "currency" field if the given value is not nil.
+func (_u *SubscriptionPlanUpdate) SetNillableCurrency(v *string) *SubscriptionPlanUpdate {
+	if v != nil {
+		_u.SetCurrency(*v)
+	}
+	return _u
+}
+
 // SetValidityDays sets the "validity_days" field.
 func (_u *SubscriptionPlanUpdate) SetValidityDays(v int) *SubscriptionPlanUpdate {
 	_u.mutation.ResetValidityDays()
@@ -277,6 +291,11 @@ func (_u *SubscriptionPlanUpdate) check() error {
 			return &ValidationError{Name: "name", err: fmt.Errorf(`ent: validator failed for field "SubscriptionPlan.name": %w`, err)}
 		}
 	}
+	if v, ok := _u.mutation.Currency(); ok {
+		if err := subscriptionplan.CurrencyValidator(v); err != nil {
+			return &ValidationError{Name: "currency", err: fmt.Errorf(`ent: validator failed for field "SubscriptionPlan.currency": %w`, err)}
+		}
+	}
 	if v, ok := _u.mutation.ValidityUnit(); ok {
 		if err := subscriptionplan.ValidityUnitValidator(v); err != nil {
 			return &ValidationError{Name: "validity_unit", err: fmt.Errorf(`ent: validator failed for field "SubscriptionPlan.validity_unit": %w`, err)}
@@ -329,6 +348,9 @@ func (_u *SubscriptionPlanUpdate) sqlSave(ctx context.Context) (_node int, err e
 	if _u.mutation.OriginalPriceCleared() {
 		_spec.ClearField(subscriptionplan.FieldOriginalPrice, field.TypeFloat64)
 	}
+	if value, ok := _u.mutation.Currency(); ok {
+		_spec.SetField(subscriptionplan.FieldCurrency, field.TypeString, value)
+	}
 	if value, ok := _u.mutation.ValidityDays(); ok {
 		_spec.SetField(subscriptionplan.FieldValidityDays, field.TypeInt, value)
 	}
@@ -473,6 +495,20 @@ func (_u *SubscriptionPlanUpdateOne) ClearOriginalPrice() *SubscriptionPlanUpdat
 	return _u
 }
 
+// SetCurrency sets the "currency" field.
+func (_u *SubscriptionPlanUpdateOne) SetCurrency(v string) *SubscriptionPlanUpdateOne {
+	_u.mutation.SetCurrency(v)
+	return _u
+}
+
+// SetNillableCurrency sets the "currency" field if the given value is not nil.
+func (_u *SubscriptionPlanUpdateOne) SetNillableCurrency(v *string) *SubscriptionPlanUpdateOne {
+	if v != nil {
+		_u.SetCurrency(*v)
+	}
+	return _u
+}
+
 // SetValidityDays sets the "validity_days" field.
 func (_u *SubscriptionPlanUpdateOne) SetValidityDays(v int) *SubscriptionPlanUpdateOne {
 	_u.mutation.ResetValidityDays()
@@ -638,6 +674,11 @@ func (_u *SubscriptionPlanUpdateOne) check() error {
 			return &ValidationError{Name: "name", err: fmt.Errorf(`ent: validator failed for field "SubscriptionPlan.name": %w`, err)}
 		}
 	}
+	if v, ok := _u.mutation.Currency(); ok {
+		if err := subscriptionplan.CurrencyValidator(v); err != nil {
+			return &ValidationError{Name: "currency", err: fmt.Errorf(`ent: validator failed for field "SubscriptionPlan.currency": %w`, err)}
+		}
+	}
 	if v, ok := _u.mutation.ValidityUnit(); ok {
 		if err := subscriptionplan.ValidityUnitValidator(v); err != nil {
 			return &ValidationError{Name: "validity_unit", err: fmt.Errorf(`ent: validator failed for field "SubscriptionPlan.validity_unit": %w`, err)}
@@ -707,6 +748,9 @@ func (_u *SubscriptionPlanUpdateOne) sqlSave(ctx context.Context) (_node *Subscr
 	if _u.mutation.OriginalPriceCleared() {
 		_spec.ClearField(subscriptionplan.FieldOriginalPrice, field.TypeFloat64)
 	}
+	if value, ok := _u.mutation.Currency(); ok {
+		_spec.SetField(subscriptionplan.FieldCurrency, field.TypeString, value)
+	}
 	if value, ok := _u.mutation.ValidityDays(); ok {
 		_spec.SetField(subscriptionplan.FieldValidityDays, field.TypeInt, value)
 	}

backend/ent/usagelog.go

@@ -75,6 +75,8 @@ type UsageLog struct {
 	ActualCost float64 `json:"actual_cost,omitempty"`
 	// RateMultiplier holds the value of the "rate_multiplier" field.
 	RateMultiplier float64 `json:"rate_multiplier,omitempty"`
+	// Whether long-context pricing changed token prices for this request
+	LongContextBillingApplied bool `json:"long_context_billing_applied,omitempty"`
 	// AccountRateMultiplier holds the value of the "account_rate_multiplier" field.
 	AccountRateMultiplier *float64 `json:"account_rate_multiplier,omitempty"`
 	// BillingType holds the value of the "billing_type" field.
@@ -196,7 +198,7 @@ func (*UsageLog) scanValues(columns []string) ([]any, error) {
 		switch columns[i] {
 		case usagelog.FieldImageSizeBreakdown:
 			values[i] = new([]byte)
-		case usagelog.FieldStream, usagelog.FieldCacheTTLOverridden:
+		case usagelog.FieldLongContextBillingApplied, usagelog.FieldStream, usagelog.FieldCacheTTLOverridden:
 			values[i] = new(sql.NullBool)
 		case usagelog.FieldInputCost, usagelog.FieldOutputCost, usagelog.FieldCacheCreationCost, usagelog.FieldCacheReadCost, usagelog.FieldTotalCost, usagelog.FieldActualCost, usagelog.FieldRateMultiplier, usagelog.FieldAccountRateMultiplier:
 			values[i] = new(sql.NullFloat64)
@@ -391,6 +393,12 @@ func (_m *UsageLog) assignValues(columns []string, values []any) error {
 			} else if value.Valid {
 				_m.RateMultiplier = value.Float64
 			}
+		case usagelog.FieldLongContextBillingApplied:
+			if value, ok := values[i].(*sql.NullBool); !ok {
+				return fmt.Errorf("unexpected type %T for field long_context_billing_applied", values[i])
+			} else if value.Valid {
+				_m.LongContextBillingApplied = value.Bool
+			}
 		case usagelog.FieldAccountRateMultiplier:
 			if value, ok := values[i].(*sql.NullFloat64); !ok {
 				return fmt.Errorf("unexpected type %T for field account_rate_multiplier", values[i])
@@ -667,6 +675,9 @@ func (_m *UsageLog) String() string {
 	builder.WriteString("rate_multiplier=")
 	builder.WriteString(fmt.Sprintf("%v", _m.RateMultiplier))
 	builder.WriteString(", ")
+	builder.WriteString("long_context_billing_applied=")
+	builder.WriteString(fmt.Sprintf("%v", _m.LongContextBillingApplied))
+	builder.WriteString(", ")
 	if v := _m.AccountRateMultiplier; v != nil {
 		builder.WriteString("account_rate_multiplier=")
 		builder.WriteString(fmt.Sprintf("%v", *v))

backend/ent/usagelog/usagelog.go

@@ -66,6 +66,8 @@ const (
 	FieldActualCost = "actual_cost"
 	// FieldRateMultiplier holds the string denoting the rate_multiplier field in the database.
 	FieldRateMultiplier = "rate_multiplier"
+	// FieldLongContextBillingApplied holds the string denoting the long_context_billing_applied field in the database.
+	FieldLongContextBillingApplied = "long_context_billing_applied"
 	// FieldAccountRateMultiplier holds the string denoting the account_rate_multiplier field in the database.
 	FieldAccountRateMultiplier = "account_rate_multiplier"
 	// FieldBillingType holds the string denoting the billing_type field in the database.
@@ -180,6 +182,7 @@ var Columns = []string{
 	FieldTotalCost,
 	FieldActualCost,
 	FieldRateMultiplier,
+	FieldLongContextBillingApplied,
 	FieldAccountRateMultiplier,
 	FieldBillingType,
 	FieldStream,
@@ -251,6 +254,8 @@ var (
 	DefaultActualCost float64
 	// DefaultRateMultiplier holds the default value on creation for the "rate_multiplier" field.
 	DefaultRateMultiplier float64
+	// DefaultLongContextBillingApplied holds the default value on creation for the "long_context_billing_applied" field.
+	DefaultLongContextBillingApplied bool
 	// DefaultBillingType holds the default value on creation for the "billing_type" field.
 	DefaultBillingType int8
 	// DefaultStream holds the default value on creation for the "stream" field.
@@ -417,6 +422,11 @@ func ByRateMultiplier(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldRateMultiplier, opts...).ToFunc()
 }
 
+// ByLongContextBillingApplied orders the results by the long_context_billing_applied field.
+func ByLongContextBillingApplied(opts ...sql.OrderTermOption) OrderOption {
+	return sql.OrderByField(FieldLongContextBillingApplied, opts...).ToFunc()
+}
+
 // ByAccountRateMultiplier orders the results by the account_rate_multiplier field.
 func ByAccountRateMultiplier(opts ...sql.OrderTermOption) OrderOption {
 	return sql.OrderByField(FieldAccountRateMultiplier, opts...).ToFunc()

backend/ent/usagelog/where.go

@@ -185,6 +185,11 @@ func RateMultiplier(v float64) predicate.UsageLog {
 	return predicate.UsageLog(sql.FieldEQ(FieldRateMultiplier, v))
 }
 
+// LongContextBillingApplied applies equality check predicate on the "long_context_billing_applied" field. It's identical to LongContextBillingAppliedEQ.
+func LongContextBillingApplied(v bool) predicate.UsageLog {
+	return predicate.UsageLog(sql.FieldEQ(FieldLongContextBillingApplied, v))
+}
+
 // AccountRateMultiplier applies equality check predicate on the "account_rate_multiplier" field. It's identical to AccountRateMultiplierEQ.
 func AccountRateMultiplier(v float64) predicate.UsageLog {
 	return predicate.UsageLog(sql.FieldEQ(FieldAccountRateMultiplier, v))
@@ -1465,6 +1470,16 @@ func RateMultiplierLTE(v float64) predicate.UsageLog {
 	return predicate.UsageLog(sql.FieldLTE(FieldRateMultiplier, v))
 }
 
+// LongContextBillingAppliedEQ applies the EQ predicate on the "long_context_billing_applied" field.
+func LongContextBillingAppliedEQ(v bool) predicate.UsageLog {
+	return predicate.UsageLog(sql.FieldEQ(FieldLongContextBillingApplied, v))
+}
+
+// LongContextBillingAppliedNEQ applies the NEQ predicate on the "long_context_billing_applied" field.
+func LongContextBillingAppliedNEQ(v bool) predicate.UsageLog {
+	return predicate.UsageLog(sql.FieldNEQ(FieldLongContextBillingApplied, v))
+}
+
 // AccountRateMultiplierEQ applies the EQ predicate on the "account_rate_multiplier" field.
 func AccountRateMultiplierEQ(v float64) predicate.UsageLog {
 	return predicate.UsageLog(sql.FieldEQ(FieldAccountRateMultiplier, v))

backend/ent/usagelog_create.go

@@ -351,6 +351,20 @@ func (_c *UsageLogCreate) SetNillableRateMultiplier(v *float64) *UsageLogCreate
 	return _c
 }
 
+// SetLongContextBillingApplied sets the "long_context_billing_applied" field.
+func (_c *UsageLogCreate) SetLongContextBillingApplied(v bool) *UsageLogCreate {
+	_c.mutation.SetLongContextBillingApplied(v)
+	return _c
+}
+
+// SetNillableLongContextBillingApplied sets the "long_context_billing_applied" field if the given value is not nil.
+func (_c *UsageLogCreate) SetNillableLongContextBillingApplied(v *bool) *UsageLogCreate {
+	if v != nil {
+		_c.SetLongContextBillingApplied(*v)
+	}
+	return _c
+}
+
 // SetAccountRateMultiplier sets the "account_rate_multiplier" field.
 func (_c *UsageLogCreate) SetAccountRateMultiplier(v float64) *UsageLogCreate {
 	_c.mutation.SetAccountRateMultiplier(v)
@@ -707,6 +721,10 @@ func (_c *UsageLogCreate) defaults() {
 		v := usagelog.DefaultRateMultiplier
 		_c.mutation.SetRateMultiplier(v)
 	}
+	if _, ok := _c.mutation.LongContextBillingApplied(); !ok {
+		v := usagelog.DefaultLongContextBillingApplied
+		_c.mutation.SetLongContextBillingApplied(v)
+	}
 	if _, ok := _c.mutation.BillingType(); !ok {
 		v := usagelog.DefaultBillingType
 		_c.mutation.SetBillingType(v)
@@ -824,6 +842,9 @@ func (_c *UsageLogCreate) check() error {
 	if _, ok := _c.mutation.RateMultiplier(); !ok {
 		return &ValidationError{Name: "rate_multiplier", err: errors.New(`ent: missing required field "UsageLog.rate_multiplier"`)}
 	}
+	if _, ok := _c.mutation.LongContextBillingApplied(); !ok {
+		return &ValidationError{Name: "long_context_billing_applied", err: errors.New(`ent: missing required field "UsageLog.long_context_billing_applied"`)}
+	}
 	if _, ok := _c.mutation.BillingType(); !ok {
 		return &ValidationError{Name: "billing_type", err: errors.New(`ent: missing required field "UsageLog.billing_type"`)}
 	}
@@ -997,6 +1018,10 @@ func (_c *UsageLogCreate) createSpec() (*UsageLog, *sqlgraph.CreateSpec) {
 		_spec.SetField(usagelog.FieldRateMultiplier, field.TypeFloat64, value)
 		_node.RateMultiplier = value
 	}
+	if value, ok := _c.mutation.LongContextBillingApplied(); ok {
+		_spec.SetField(usagelog.FieldLongContextBillingApplied, field.TypeBool, value)
+		_node.LongContextBillingApplied = value
+	}
 	if value, ok := _c.mutation.AccountRateMultiplier(); ok {
 		_spec.SetField(usagelog.FieldAccountRateMultiplier, field.TypeFloat64, value)
 		_node.AccountRateMultiplier = &value
@@ -1650,6 +1675,18 @@ func (u *UsageLogUpsert) AddRateMultiplier(v float64) *UsageLogUpsert {
 	return u
 }
 
+// SetLongContextBillingApplied sets the "long_context_billing_applied" field.
+func (u *UsageLogUpsert) SetLongContextBillingApplied(v bool) *UsageLogUpsert {
+	u.Set(usagelog.FieldLongContextBillingApplied, v)
+	return u
+}
+
+// UpdateLongContextBillingApplied sets the "long_context_billing_applied" field to the value that was provided on create.
+func (u *UsageLogUpsert) UpdateLongContextBillingApplied() *UsageLogUpsert {
+	u.SetExcluded(usagelog.FieldLongContextBillingApplied)
+	return u
+}
+
 // SetAccountRateMultiplier sets the "account_rate_multiplier" field.
 func (u *UsageLogUpsert) SetAccountRateMultiplier(v float64) *UsageLogUpsert {
 	u.Set(usagelog.FieldAccountRateMultiplier, v)
@@ -2531,6 +2568,20 @@ func (u *UsageLogUpsertOne) UpdateRateMultiplier() *UsageLogUpsertOne {
 	})
 }
 
+// SetLongContextBillingApplied sets the "long_context_billing_applied" field.
+func (u *UsageLogUpsertOne) SetLongContextBillingApplied(v bool) *UsageLogUpsertOne {
+	return u.Update(func(s *UsageLogUpsert) {
+		s.SetLongContextBillingApplied(v)
+	})
+}
+
+// UpdateLongContextBillingApplied sets the "long_context_billing_applied" field to the value that was provided on create.
+func (u *UsageLogUpsertOne) UpdateLongContextBillingApplied() *UsageLogUpsertOne {
+	return u.Update(func(s *UsageLogUpsert) {
+		s.UpdateLongContextBillingApplied()
+	})
+}
+
 // SetAccountRateMultiplier sets the "account_rate_multiplier" field.
 func (u *UsageLogUpsertOne) SetAccountRateMultiplier(v float64) *UsageLogUpsertOne {
 	return u.Update(func(s *UsageLogUpsert) {
@@ -3631,6 +3682,20 @@ func (u *UsageLogUpsertBulk) UpdateRateMultiplier() *UsageLogUpsertBulk {
 	})
 }
 
+// SetLongContextBillingApplied sets the "long_context_billing_applied" field.
+func (u *UsageLogUpsertBulk) SetLongContextBillingApplied(v bool) *UsageLogUpsertBulk {
+	return u.Update(func(s *UsageLogUpsert) {
+		s.SetLongContextBillingApplied(v)
+	})
+}
+
+// UpdateLongContextBillingApplied sets the "long_context_billing_applied" field to the value that was provided on create.
+func (u *UsageLogUpsertBulk) UpdateLongContextBillingApplied() *UsageLogUpsertBulk {
+	return u.Update(func(s *UsageLogUpsert) {
+		s.UpdateLongContextBillingApplied()
+	})
+}
+
 // SetAccountRateMultiplier sets the "account_rate_multiplier" field.
 func (u *UsageLogUpsertBulk) SetAccountRateMultiplier(v float64) *UsageLogUpsertBulk {
 	return u.Update(func(s *UsageLogUpsert) {

backend/ent/usagelog_update.go

@@ -542,6 +542,20 @@ func (_u *UsageLogUpdate) AddRateMultiplier(v float64) *UsageLogUpdate {
 	return _u
 }
 
+// SetLongContextBillingApplied sets the "long_context_billing_applied" field.
+func (_u *UsageLogUpdate) SetLongContextBillingApplied(v bool) *UsageLogUpdate {
+	_u.mutation.SetLongContextBillingApplied(v)
+	return _u
+}
+
+// SetNillableLongContextBillingApplied sets the "long_context_billing_applied" field if the given value is not nil.
+func (_u *UsageLogUpdate) SetNillableLongContextBillingApplied(v *bool) *UsageLogUpdate {
+	if v != nil {
+		_u.SetLongContextBillingApplied(*v)
+	}
+	return _u
+}
+
 // SetAccountRateMultiplier sets the "account_rate_multiplier" field.
 func (_u *UsageLogUpdate) SetAccountRateMultiplier(v float64) *UsageLogUpdate {
 	_u.mutation.ResetAccountRateMultiplier()
@@ -1199,6 +1213,9 @@ func (_u *UsageLogUpdate) sqlSave(ctx context.Context) (_node int, err error) {
 	if value, ok := _u.mutation.AddedRateMultiplier(); ok {
 		_spec.AddField(usagelog.FieldRateMultiplier, field.TypeFloat64, value)
 	}
+	if value, ok := _u.mutation.LongContextBillingApplied(); ok {
+		_spec.SetField(usagelog.FieldLongContextBillingApplied, field.TypeBool, value)
+	}
 	if value, ok := _u.mutation.AccountRateMultiplier(); ok {
 		_spec.SetField(usagelog.FieldAccountRateMultiplier, field.TypeFloat64, value)
 	}
@@ -1982,6 +1999,20 @@ func (_u *UsageLogUpdateOne) AddRateMultiplier(v float64) *UsageLogUpdateOne {
 	return _u
 }
 
+// SetLongContextBillingApplied sets the "long_context_billing_applied" field.
+func (_u *UsageLogUpdateOne) SetLongContextBillingApplied(v bool) *UsageLogUpdateOne {
+	_u.mutation.SetLongContextBillingApplied(v)
+	return _u
+}
+
+// SetNillableLongContextBillingApplied sets the "long_context_billing_applied" field if the given value is not nil.
+func (_u *UsageLogUpdateOne) SetNillableLongContextBillingApplied(v *bool) *UsageLogUpdateOne {
+	if v != nil {
+		_u.SetLongContextBillingApplied(*v)
+	}
+	return _u
+}
+
 // SetAccountRateMultiplier sets the "account_rate_multiplier" field.
 func (_u *UsageLogUpdateOne) SetAccountRateMultiplier(v float64) *UsageLogUpdateOne {
 	_u.mutation.ResetAccountRateMultiplier()
@@ -2669,6 +2700,9 @@ func (_u *UsageLogUpdateOne) sqlSave(ctx context.Context) (_node *UsageLog, err
 	if value, ok := _u.mutation.AddedRateMultiplier(); ok {
 		_spec.AddField(usagelog.FieldRateMultiplier, field.TypeFloat64, value)
 	}
+	if value, ok := _u.mutation.LongContextBillingApplied(); ok {
+		_spec.SetField(usagelog.FieldLongContextBillingApplied, field.TypeBool, value)
+	}
 	if value, ok := _u.mutation.AccountRateMultiplier(); ok {
 		_spec.SetField(usagelog.FieldAccountRateMultiplier, field.TypeFloat64, value)
 	}

backend/go.mod

@@ -44,6 +44,7 @@ require (
 	go.uber.org/zap v1.24.0
 	golang.org/x/crypto v0.51.0
 	golang.org/x/image v0.39.0
+	golang.org/x/mod v0.35.0
 	golang.org/x/net v0.55.0
 	golang.org/x/sync v0.20.0
 	golang.org/x/term v0.43.0
@@ -176,7 +177,6 @@ require (
 	go.uber.org/multierr v1.9.0 // indirect
 	golang.org/x/arch v0.3.0 // indirect
 	golang.org/x/exp v0.0.0-20251023183803-a4bb9ffd2546 // indirect
-	golang.org/x/mod v0.35.0 // indirect
 	golang.org/x/sys v0.45.0 // indirect
 	golang.org/x/text v0.37.0 // indirect
 	golang.org/x/tools v0.44.0 // indirect

backend/go.sum

@@ -166,6 +166,8 @@ github.com/google/go-querystring v1.1.0/go.mod h1:Kcdr2DB4koayq7X8pmAG4sNG59So17
 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=
@@ -220,6 +222,8 @@ 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=
@@ -253,6 +257,8 @@ 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=
@@ -282,6 +288,8 @@ github.com/refraction-networking/utls v1.8.2 h1:j4Q1gJj0xngdeH+Ox/qND11aEfhpgoEv
 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=
@@ -314,6 +322,8 @@ 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=

backend/internal/config/config.go

@@ -6,6 +6,7 @@ import (
 	"encoding/hex"
 	"fmt"
 	"log/slog"
+	"math"
 	"net/url"
 	"os"
 	"strings"
@@ -94,6 +95,7 @@ type Config struct {
 	Update                  UpdateConfig                  `mapstructure:"update"`
 	Idempotency             IdempotencyConfig             `mapstructure:"idempotency"`
 	BatchImage              BatchImageConfig              `mapstructure:"batch_image"`
+	ImageStorage            ImageStorageConfig            `mapstructure:"image_storage"`
 }
 
 type LogConfig struct {
@@ -226,6 +228,33 @@ type BatchImageConfig struct {
 	VertexGCSBaseURL             string `mapstructure:"vertex_gcs_base_url"`
 }
 
+// ImageStorageConfig 配置异步图片任务结果上传的 S3 兼容对象存储。
+// Enabled 同时作为异步图片任务功能的总开关：未启用或未配置完整凭证时，
+// 异步生图接口整体禁用，避免把上游返回的大 base64 结果塞进 Redis。
+type ImageStorageConfig struct {
+	Enabled         bool   `mapstructure:"enabled"`
+	Endpoint        string `mapstructure:"endpoint"` // e.g. https://<account_id>.r2.cloudflarestorage.com
+	Region          string `mapstructure:"region"`   // R2 用 "auto"
+	Bucket          string `mapstructure:"bucket"`
+	AccessKeyID     string `mapstructure:"access_key_id"`
+	SecretAccessKey string `mapstructure:"secret_access_key"`
+	Prefix          string `mapstructure:"prefix"`               // S3 key 前缀，如 "images/"
+	ForcePathStyle  bool   `mapstructure:"force_path_style"`     // MinIO/路径风格桶
+	PublicBaseURL   string `mapstructure:"public_base_url"`      // 配了则返回 public_base_url/key 直链；否则 presigned
+	PresignExpiry   int    `mapstructure:"presign_expiry_hours"` // public_base_url 为空时的 presigned 过期时长(小时)
+	MaxDownloadByte int64  `mapstructure:"max_download_bytes"`   // 下载上游 url 图片的字节上限
+}
+
+// IsConfigured 检查对象存储必要字段是否已配置
+func (c *ImageStorageConfig) IsConfigured() bool {
+	return c.Bucket != "" && c.AccessKeyID != "" && c.SecretAccessKey != ""
+}
+
+// Active 返回异步图片任务是否可用：开关打开且凭证齐全
+func (c *ImageStorageConfig) Active() bool {
+	return c.Enabled && c.IsConfigured()
+}
+
 type LinuxDoConnectConfig struct {
 	Enabled             bool   `mapstructure:"enabled"`
 	ClientID            string `mapstructure:"client_id"`
@@ -580,6 +609,18 @@ type TokenRefreshConfig struct {
 	MaxRetries int `mapstructure:"max_retries"`
 	// 重试退避基础时间（秒）
 	RetryBackoffSeconds int `mapstructure:"retry_backoff_seconds"`
+	// 每次从数据库读取的候选账号上限
+	CandidatePageSize int `mapstructure:"candidate_page_size"`
+	// 每个平台允许的并发刷新数
+	ProviderConcurrency int `mapstructure:"provider_concurrency"`
+	// 每个平台、每个进程允许的刷新请求速率
+	ProviderQPS int `mapstructure:"provider_qps"`
+	// 一个周期内连续临时失败达到此值后停止该平台
+	ProviderFailureThreshold int `mapstructure:"provider_failure_threshold"`
+	// 单次上游刷新尝试的超时（秒）
+	AttemptTimeoutSeconds int `mapstructure:"attempt_timeout_seconds"`
+	// 单个后台刷新周期的总超时（秒）
+	CycleTimeoutSeconds int `mapstructure:"cycle_timeout_seconds"`
 }
 
 type PricingConfig struct {
@@ -601,6 +642,7 @@ type ServerConfig struct {
 	Host               string    `mapstructure:"host"`
 	Port               int       `mapstructure:"port"`
 	Mode               string    `mapstructure:"mode"`                  // debug/release
+	EnableServerTiming bool      `mapstructure:"enable_server_timing"`  // Admin UI Server-Timing response header
 	FrontendURL        string    `mapstructure:"frontend_url"`          // 前端基础 URL，用于生成邮件中的外部链接
 	ReadHeaderTimeout  int       `mapstructure:"read_header_timeout"`   // 读取请求头超时（秒）
 	IdleTimeout        int       `mapstructure:"idle_timeout"`          // 空闲连接超时（秒）
@@ -747,6 +789,11 @@ type GatewayConfig struct {
 	// OpenAIResponseHeaderTimeout: OpenAI/Codex 上游等待响应头的超时时间（秒），0表示无超时
 	// OpenAI/Codex 请求可能在上游排队较久；默认不使用通用响应头超时截断。
 	OpenAIResponseHeaderTimeout int `mapstructure:"openai_response_header_timeout"`
+	// OpenAIFirstOutputTimeoutSeconds: native HTTP Responses 首个语义输出超时（秒），0表示禁用。
+	OpenAIFirstOutputTimeoutSeconds int `mapstructure:"openai_first_output_timeout_seconds"`
+	// OpenAIHighEffortFirstOutputTimeoutSeconds: high/xhigh/max 推理的首个语义输出超时（秒）。
+	// 0 表示回退到 OpenAIFirstOutputTimeoutSeconds。
+	OpenAIHighEffortFirstOutputTimeoutSeconds int `mapstructure:"openai_high_effort_first_output_timeout_seconds"`
 	// 请求体最大字节数，用于网关请求体大小限制
 	MaxBodySize int64 `mapstructure:"max_body_size"`
 	// 非流式上游响应体读取上限（字节），用于防止无界读取导致内存放大
@@ -818,6 +865,8 @@ type GatewayConfig struct {
 	ImageStreamDataIntervalTimeout int `mapstructure:"image_stream_data_interval_timeout"`
 	// ImageStreamKeepaliveInterval: 图片流式 keepalive 间隔（秒），0表示禁用
 	ImageStreamKeepaliveInterval int `mapstructure:"image_stream_keepalive_interval"`
+	// ImageNonstreamKeepaliveInterval: 图片非流式 JSON keepalive 间隔（秒），0表示禁用
+	ImageNonstreamKeepaliveInterval int `mapstructure:"image_nonstream_keepalive_interval"`
 	// MaxLineSize: 上游 SSE 单行最大字节数（0使用默认值）
 	MaxLineSize int `mapstructure:"max_line_size"`
 
@@ -916,6 +965,9 @@ func (c *UserMessageQueueConfig) GetEffectiveMode() string {
 	return ""
 }
 
+// DefaultOpenAIWSClientFirstMessageTimeoutSeconds preserves the legacy ingress deadline.
+const DefaultOpenAIWSClientFirstMessageTimeoutSeconds = 30
+
 // GatewayOpenAIWSConfig OpenAI Responses WebSocket 配置。
 // 注意：默认全局开启；如需回滚可使用 force_http 或关闭 enabled。
 type GatewayOpenAIWSConfig struct {
@@ -923,6 +975,15 @@ type GatewayOpenAIWSConfig struct {
 	ModeRouterV2Enabled bool `mapstructure:"mode_router_v2_enabled"`
 	// IngressModeDefault: ingress 默认模式（off/ctx_pool/passthrough/http_bridge）
 	IngressModeDefault string `mapstructure:"ingress_mode_default"`
+	// ClientFirstMessageTimeoutSeconds bounds the total time to read and decompress
+	// the first client response.create message after the WebSocket upgrade.
+	ClientFirstMessageTimeoutSeconds int `mapstructure:"client_first_message_timeout_seconds"`
+	// IngressInterTurnIdleTimeoutSeconds bounds the time a client may remain idle
+	// between completed ingress turns. Zero disables this protection.
+	IngressInterTurnIdleTimeoutSeconds int `mapstructure:"ingress_inter_turn_idle_timeout_seconds"`
+	// MaxIngressConnectionsPerAPIKey bounds live client WebSocket ingress sessions
+	// per API key across all instances. Zero disables this protection.
+	MaxIngressConnectionsPerAPIKey int `mapstructure:"max_ingress_connections_per_api_key"`
 	// Enabled: 全局总开关（默认 true）
 	Enabled bool `mapstructure:"enabled"`
 	// OAuthEnabled: 是否允许 OpenAI OAuth 账号使用 WS
@@ -1020,11 +1081,35 @@ type GatewayOpenAIWSSchedulerScoreWeights struct {
 	Reset float64 `mapstructure:"reset"`
 	// QuotaHeadroom 倾向 7d 剩余额度更健康的账号；默认 0（关闭，不改变原有行为）。
 	QuotaHeadroom float64 `mapstructure:"quota_headroom"`
+	// UpstreamCost 倾向上游声明倍率更低的账号；默认 0（关闭，不改变原有行为）。
+	UpstreamCost float64 `mapstructure:"upstream_cost"`
 	// PreviousResponse/SessionSticky 仅在开启 OpenAI 高级调度的粘性加权时生效。
 	PreviousResponse float64 `mapstructure:"previous_response"`
 	SessionSticky    float64 `mapstructure:"session_sticky"`
 }
 
+func (w GatewayOpenAIWSSchedulerScoreWeights) BaseWeightSum() float64 {
+	return w.Priority + w.Load + w.Queue + w.ErrorRate + w.TTFT + w.Reset + w.QuotaHeadroom + w.UpstreamCost
+}
+
+func (w GatewayOpenAIWSSchedulerScoreWeights) TotalWeightSum() float64 {
+	return w.BaseWeightSum() + w.PreviousResponse + w.SessionSticky
+}
+
+func (w GatewayOpenAIWSSchedulerScoreWeights) IsValid() bool {
+	for _, weight := range []float64{
+		w.Priority, w.Load, w.Queue, w.ErrorRate, w.TTFT, w.Reset,
+		w.QuotaHeadroom, w.UpstreamCost, w.PreviousResponse, w.SessionSticky,
+	} {
+		if weight < 0 || math.IsNaN(weight) || math.IsInf(weight, 0) {
+			return false
+		}
+	}
+	baseSum := w.BaseWeightSum()
+	return baseSum > 0 && !math.IsNaN(baseSum) && !math.IsInf(baseSum, 0) &&
+		!math.IsNaN(w.TotalWeightSum()) && !math.IsInf(w.TotalWeightSum(), 0)
+}
+
 // GatewayOpenAISchedulerConfig OpenAI 高级调度器配置。
 type GatewayOpenAISchedulerConfig struct {
 	// StickyEscapeEnabled: 是否允许 session_hash sticky 在账号健康度劣化时临时逃逸
@@ -1453,6 +1538,9 @@ func load(allowMissingJWTSecret bool) (*Config, error) {
 	// 环境变量支持
 	viper.AutomaticEnv()
 	viper.SetEnvKeyReplacer(strings.NewReplacer(".", "_"))
+	if err := viper.BindEnv("server.enable_server_timing", "ENABLE_SERVER_TIMING"); err != nil {
+		return nil, fmt.Errorf("bind ENABLE_SERVER_TIMING: %w", err)
+	}
 
 	// 默认值
 	setDefaults()
@@ -1608,6 +1696,7 @@ func setDefaults() {
 	viper.SetDefault("server.host", "0.0.0.0")
 	viper.SetDefault("server.port", 8080)
 	viper.SetDefault("server.mode", "release")
+	viper.SetDefault("server.enable_server_timing", false)
 	viper.SetDefault("server.frontend_url", "")
 	viper.SetDefault("server.read_header_timeout", 30) // 30秒读取请求头
 	viper.SetDefault("server.idle_timeout", 120)       // 120秒空闲超时
@@ -1830,6 +1919,14 @@ func setDefaults() {
 	viper.SetDefault("batch_image.vertex_batch_prediction_base_url", "")
 	viper.SetDefault("batch_image.vertex_gcs_base_url", "")
 
+	// Image storage (async image task result offload to S3-compatible object storage)
+	viper.SetDefault("image_storage.enabled", false)
+	viper.SetDefault("image_storage.region", "auto")
+	viper.SetDefault("image_storage.prefix", "images/")
+	viper.SetDefault("image_storage.force_path_style", false)
+	viper.SetDefault("image_storage.presign_expiry_hours", 24)
+	viper.SetDefault("image_storage.max_download_bytes", 33554432)
+
 	// Ops (vNext)
 	viper.SetDefault("ops.enabled", true)
 	viper.SetDefault("ops.use_preaggregated_tables", true)
@@ -1931,6 +2028,8 @@ func setDefaults() {
 	// Gateway
 	viper.SetDefault("gateway.response_header_timeout", 600) // 600秒(10分钟)等待上游响应头，LLM高负载时可能排队较久
 	viper.SetDefault("gateway.openai_response_header_timeout", 0)
+	viper.SetDefault("gateway.openai_first_output_timeout_seconds", 0)
+	viper.SetDefault("gateway.openai_high_effort_first_output_timeout_seconds", 0)
 	viper.SetDefault("gateway.log_upstream_error_body", true)
 	viper.SetDefault("gateway.log_upstream_error_body_max_bytes", 2048)
 	viper.SetDefault("gateway.inject_beta_for_apikey", false)
@@ -1945,6 +2044,9 @@ func setDefaults() {
 	viper.SetDefault("gateway.openai_ws.enabled", true)
 	viper.SetDefault("gateway.openai_ws.mode_router_v2_enabled", false)
 	viper.SetDefault("gateway.openai_ws.ingress_mode_default", "ctx_pool")
+	viper.SetDefault("gateway.openai_ws.client_first_message_timeout_seconds", DefaultOpenAIWSClientFirstMessageTimeoutSeconds)
+	viper.SetDefault("gateway.openai_ws.ingress_inter_turn_idle_timeout_seconds", 300)
+	viper.SetDefault("gateway.openai_ws.max_ingress_connections_per_api_key", 64)
 	viper.SetDefault("gateway.openai_ws.oauth_enabled", true)
 	viper.SetDefault("gateway.openai_ws.apikey_enabled", true)
 	viper.SetDefault("gateway.openai_ws.force_http", false)
@@ -1992,6 +2094,7 @@ func setDefaults() {
 	viper.SetDefault("gateway.openai_ws.scheduler_score_weights.ttft", 0.5)
 	viper.SetDefault("gateway.openai_ws.scheduler_score_weights.reset", 0.0)
 	viper.SetDefault("gateway.openai_ws.scheduler_score_weights.quota_headroom", 0.0)
+	viper.SetDefault("gateway.openai_ws.scheduler_score_weights.upstream_cost", 0.0)
 	viper.SetDefault("gateway.openai_ws.scheduler_score_weights.previous_response", 5.0)
 	viper.SetDefault("gateway.openai_ws.scheduler_score_weights.session_sticky", 3.0)
 	// OpenAI HTTP upstream protocol strategy
@@ -2024,6 +2127,7 @@ func setDefaults() {
 	viper.SetDefault("gateway.stream_keepalive_interval", 10)
 	viper.SetDefault("gateway.image_stream_data_interval_timeout", 900)
 	viper.SetDefault("gateway.image_stream_keepalive_interval", 10)
+	viper.SetDefault("gateway.image_nonstream_keepalive_interval", 0)
 	viper.SetDefault("gateway.max_line_size", 500*1024*1024)
 	viper.SetDefault("gateway.scheduling.sticky_session_max_waiting", 3)
 	viper.SetDefault("gateway.scheduling.sticky_session_wait_timeout", 120*time.Second)
@@ -2082,6 +2186,12 @@ func setDefaults() {
 	viper.SetDefault("token_refresh.refresh_before_expiry_hours", 0.5) // 提前30分钟刷新（适配Google 1小时token）
 	viper.SetDefault("token_refresh.max_retries", 3)                   // 最多重试3次
 	viper.SetDefault("token_refresh.retry_backoff_seconds", 2)         // 重试退避基础2秒
+	viper.SetDefault("token_refresh.candidate_page_size", 200)
+	viper.SetDefault("token_refresh.provider_concurrency", 4)
+	viper.SetDefault("token_refresh.provider_qps", 2)
+	viper.SetDefault("token_refresh.provider_failure_threshold", 3)
+	viper.SetDefault("token_refresh.attempt_timeout_seconds", 15)
+	viper.SetDefault("token_refresh.cycle_timeout_seconds", 240)
 
 	// Gemini OAuth - configure via environment variables or config file
 	// GEMINI_OAUTH_CLIENT_ID and GEMINI_OAUTH_CLIENT_SECRET
@@ -2635,6 +2745,14 @@ func (c *Config) Validate() error {
 	if c.Gateway.OpenAIResponseHeaderTimeout < 0 {
 		return fmt.Errorf("gateway.openai_response_header_timeout must be non-negative")
 	}
+	if c.Gateway.OpenAIFirstOutputTimeoutSeconds < 0 || c.Gateway.OpenAIFirstOutputTimeoutSeconds > 600 ||
+		(c.Gateway.OpenAIFirstOutputTimeoutSeconds > 0 && c.Gateway.OpenAIFirstOutputTimeoutSeconds < 30) {
+		return fmt.Errorf("gateway.openai_first_output_timeout_seconds must be 0 or between 30-600 seconds")
+	}
+	if c.Gateway.OpenAIHighEffortFirstOutputTimeoutSeconds < 0 || c.Gateway.OpenAIHighEffortFirstOutputTimeoutSeconds > 1800 ||
+		(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 strings.TrimSpace(c.Gateway.ConnectionPoolIsolation) != "" {
 		switch c.Gateway.ConnectionPoolIsolation {
 		case ConnectionPoolIsolationProxy, ConnectionPoolIsolationAccount, ConnectionPoolIsolationAccountProxy:
@@ -2710,6 +2828,13 @@ func (c *Config) Validate() error {
 		(c.Gateway.ImageStreamKeepaliveInterval < 5 || c.Gateway.ImageStreamKeepaliveInterval > 60) {
 		return fmt.Errorf("gateway.image_stream_keepalive_interval must be 0 or between 5-60 seconds")
 	}
+	if c.Gateway.ImageNonstreamKeepaliveInterval < 0 {
+		return fmt.Errorf("gateway.image_nonstream_keepalive_interval must be non-negative")
+	}
+	if c.Gateway.ImageNonstreamKeepaliveInterval != 0 &&
+		(c.Gateway.ImageNonstreamKeepaliveInterval < 5 || c.Gateway.ImageNonstreamKeepaliveInterval > 60) {
+		return fmt.Errorf("gateway.image_nonstream_keepalive_interval must be 0 or between 5-60 seconds")
+	}
 	// 兼容旧键 sticky_previous_response_ttl_seconds
 	if c.Gateway.OpenAIWS.StickyResponseIDTTLSeconds <= 0 && c.Gateway.OpenAIWS.StickyPreviousResponseTTLSeconds > 0 {
 		c.Gateway.OpenAIWS.StickyResponseIDTTLSeconds = c.Gateway.OpenAIWS.StickyPreviousResponseTTLSeconds
@@ -2717,6 +2842,15 @@ func (c *Config) Validate() error {
 	if c.Gateway.OpenAIWS.MaxConnsPerAccount <= 0 {
 		return fmt.Errorf("gateway.openai_ws.max_conns_per_account must be positive")
 	}
+	if c.Gateway.OpenAIWS.ClientFirstMessageTimeoutSeconds <= 0 {
+		return fmt.Errorf("gateway.openai_ws.client_first_message_timeout_seconds must be positive")
+	}
+	if c.Gateway.OpenAIWS.IngressInterTurnIdleTimeoutSeconds < 0 {
+		return fmt.Errorf("gateway.openai_ws.ingress_inter_turn_idle_timeout_seconds must be non-negative")
+	}
+	if c.Gateway.OpenAIWS.MaxIngressConnectionsPerAPIKey < 0 {
+		return fmt.Errorf("gateway.openai_ws.max_ingress_connections_per_api_key must be non-negative")
+	}
 	if c.Gateway.OpenAIWS.MinIdlePerAccount < 0 {
 		return fmt.Errorf("gateway.openai_ws.min_idle_per_account must be non-negative")
 	}
@@ -2827,25 +2961,26 @@ func (c *Config) Validate() error {
 	if c.Gateway.OpenAIHTTP2.FallbackTTLSeconds < 0 {
 		return fmt.Errorf("gateway.openai_http2.fallback_ttl_seconds must be non-negative")
 	}
-	if c.Gateway.OpenAIWS.SchedulerScoreWeights.Priority < 0 ||
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.Load < 0 ||
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.Queue < 0 ||
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.ErrorRate < 0 ||
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.TTFT < 0 ||
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.QuotaHeadroom < 0 ||
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.PreviousResponse < 0 ||
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.SessionSticky < 0 {
-		return fmt.Errorf("gateway.openai_ws.scheduler_score_weights.* must be non-negative")
+	weights := c.Gateway.OpenAIWS.SchedulerScoreWeights
+	for _, weight := range []float64{
+		weights.Priority, weights.Load, weights.Queue, weights.ErrorRate, weights.TTFT,
+		weights.Reset, weights.QuotaHeadroom, weights.UpstreamCost,
+		weights.PreviousResponse, weights.SessionSticky,
+	} {
+		if weight < 0 || math.IsNaN(weight) || math.IsInf(weight, 0) {
+			return fmt.Errorf("gateway.openai_ws.scheduler_score_weights.* must be non-negative and finite")
+		}
 	}
-	weightSum := c.Gateway.OpenAIWS.SchedulerScoreWeights.Priority +
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.Load +
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.Queue +
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.ErrorRate +
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.TTFT +
-		c.Gateway.OpenAIWS.SchedulerScoreWeights.QuotaHeadroom
+	weightSum := weights.BaseWeightSum()
 	if weightSum <= 0 {
 		return fmt.Errorf("gateway.openai_ws.scheduler_score_weights must not all be zero")
 	}
+	if math.IsNaN(weightSum) || math.IsInf(weightSum, 0) {
+		return fmt.Errorf("gateway.openai_ws.scheduler_score_weights base-weight sum must be finite")
+	}
+	if totalWeightSum := weights.TotalWeightSum(); math.IsNaN(totalWeightSum) || math.IsInf(totalWeightSum, 0) {
+		return fmt.Errorf("gateway.openai_ws.scheduler_score_weights total-weight sum must be finite")
+	}
 	if c.Gateway.OpenAIScheduler.StickyEscapeTTFTMs <= 0 {
 		return fmt.Errorf("gateway.openai_scheduler.sticky_escape_ttft_ms must be positive")
 	}

backend/internal/config/config_test.go

@@ -1,6 +1,7 @@
 package config
 
 import (
+	"math"
 	"os"
 	"path/filepath"
 	"strings"
@@ -17,6 +18,23 @@ func resetViperWithJWTSecret(t *testing.T) {
 	t.Setenv("JWT_SECRET", strings.Repeat("x", 32))
 }
 
+func TestLoadServerTimingConfig(t *testing.T) {
+	t.Run("disabled by default", func(t *testing.T) {
+		resetViperWithJWTSecret(t)
+		cfg, err := Load()
+		require.NoError(t, err)
+		require.False(t, cfg.Server.EnableServerTiming)
+	})
+
+	t.Run("enabled by exact environment variable", func(t *testing.T) {
+		resetViperWithJWTSecret(t)
+		t.Setenv("ENABLE_SERVER_TIMING", "true")
+		cfg, err := Load()
+		require.NoError(t, err)
+		require.True(t, cfg.Server.EnableServerTiming)
+	})
+}
+
 func TestLoadForBootstrapAllowsMissingJWTSecret(t *testing.T) {
 	viper.Reset()
 	t.Setenv("JWT_SECRET", "")
@@ -81,6 +99,34 @@ func TestLoadDefaultSchedulingConfig(t *testing.T) {
 	}
 }
 
+func TestLoadDefaultOpenAIFirstOutputTimeoutsDisabled(t *testing.T) {
+	resetViperWithJWTSecret(t)
+
+	cfg, err := Load()
+	require.NoError(t, err)
+	require.Zero(t, cfg.Gateway.OpenAIFirstOutputTimeoutSeconds)
+	require.Zero(t, cfg.Gateway.OpenAIHighEffortFirstOutputTimeoutSeconds)
+}
+
+func TestLoadOpenAIFirstOutputTimeoutsFromEnv(t *testing.T) {
+	resetViperWithJWTSecret(t)
+	t.Setenv("GATEWAY_OPENAI_FIRST_OUTPUT_TIMEOUT_SECONDS", "90")
+	t.Setenv("GATEWAY_OPENAI_HIGH_EFFORT_FIRST_OUTPUT_TIMEOUT_SECONDS", "240")
+
+	cfg, err := Load()
+	require.NoError(t, err)
+	require.Equal(t, 90, cfg.Gateway.OpenAIFirstOutputTimeoutSeconds)
+	require.Equal(t, 240, cfg.Gateway.OpenAIHighEffortFirstOutputTimeoutSeconds)
+}
+
+func TestValidateOpenAIFirstOutputTimeoutMinimum(t *testing.T) {
+	resetViperWithJWTSecret(t)
+	cfg, err := Load()
+	require.NoError(t, err)
+	cfg.Gateway.OpenAIFirstOutputTimeoutSeconds = 30
+	require.NoError(t, cfg.Validate())
+}
+
 func TestLoadDefaultOpenAIWSConfig(t *testing.T) {
 	resetViperWithJWTSecret(t)
 
@@ -170,6 +216,9 @@ func TestLoadDefaultOpenAIWSConfig(t *testing.T) {
 	if cfg.Gateway.OpenAIWS.SchedulerScoreWeights.QuotaHeadroom != 0 {
 		t.Fatalf("Gateway.OpenAIWS.SchedulerScoreWeights.QuotaHeadroom = %v, want 0", cfg.Gateway.OpenAIWS.SchedulerScoreWeights.QuotaHeadroom)
 	}
+	if cfg.Gateway.OpenAIWS.SchedulerScoreWeights.UpstreamCost != 0 {
+		t.Fatalf("Gateway.OpenAIWS.SchedulerScoreWeights.UpstreamCost = %v, want 0", cfg.Gateway.OpenAIWS.SchedulerScoreWeights.UpstreamCost)
+	}
 	if !cfg.Gateway.OpenAIWS.StoreDisabledForceNewConn {
 		t.Fatalf("Gateway.OpenAIWS.StoreDisabledForceNewConn = false, want true")
 	}
@@ -182,6 +231,28 @@ func TestLoadDefaultOpenAIWSConfig(t *testing.T) {
 	if cfg.Gateway.OpenAIWS.IngressModeDefault != "ctx_pool" {
 		t.Fatalf("Gateway.OpenAIWS.IngressModeDefault = %q, want %q", cfg.Gateway.OpenAIWS.IngressModeDefault, "ctx_pool")
 	}
+	if cfg.Gateway.OpenAIWS.ClientFirstMessageTimeoutSeconds != DefaultOpenAIWSClientFirstMessageTimeoutSeconds {
+		t.Fatalf(
+			"Gateway.OpenAIWS.ClientFirstMessageTimeoutSeconds = %d, want %d",
+			cfg.Gateway.OpenAIWS.ClientFirstMessageTimeoutSeconds,
+			DefaultOpenAIWSClientFirstMessageTimeoutSeconds,
+		)
+	}
+	if cfg.Gateway.OpenAIWS.IngressInterTurnIdleTimeoutSeconds != 300 {
+		t.Fatalf("Gateway.OpenAIWS.IngressInterTurnIdleTimeoutSeconds = %d, want 300", cfg.Gateway.OpenAIWS.IngressInterTurnIdleTimeoutSeconds)
+	}
+	if cfg.Gateway.OpenAIWS.MaxIngressConnectionsPerAPIKey != 64 {
+		t.Fatalf("Gateway.OpenAIWS.MaxIngressConnectionsPerAPIKey = %d, want 64", cfg.Gateway.OpenAIWS.MaxIngressConnectionsPerAPIKey)
+	}
+}
+
+func TestLoadOpenAIWSClientFirstMessageTimeoutFromEnv(t *testing.T) {
+	resetViperWithJWTSecret(t)
+	t.Setenv("GATEWAY_OPENAI_WS_CLIENT_FIRST_MESSAGE_TIMEOUT_SECONDS", "120")
+
+	cfg, err := Load()
+	require.NoError(t, err)
+	require.Equal(t, 120, cfg.Gateway.OpenAIWS.ClientFirstMessageTimeoutSeconds)
 }
 
 func TestLoadDefaultOpenAICompactModel(t *testing.T) {
@@ -236,6 +307,15 @@ func TestLoadOpenAIResponseHeaderTimeoutFromEnv(t *testing.T) {
 	require.Equal(t, 1800, cfg.Gateway.OpenAIResponseHeaderTimeout)
 }
 
+func TestLoadImageNonstreamKeepaliveFromEnv(t *testing.T) {
+	resetViperWithJWTSecret(t)
+	t.Setenv("GATEWAY_IMAGE_NONSTREAM_KEEPALIVE_INTERVAL", "15")
+
+	cfg, err := Load()
+	require.NoError(t, err)
+	require.Equal(t, 15, cfg.Gateway.ImageNonstreamKeepaliveInterval)
+}
+
 func TestLoadOpenAIWSStickyTTLCompatibility(t *testing.T) {
 	resetViperWithJWTSecret(t)
 	t.Setenv("GATEWAY_OPENAI_WS_STICKY_RESPONSE_ID_TTL_SECONDS", "0")
@@ -1316,6 +1396,16 @@ func TestValidateConfigErrors(t *testing.T) {
 			mutate:  func(c *Config) { c.Gateway.OpenAIResponseHeaderTimeout = -1 },
 			wantErr: "gateway.openai_response_header_timeout",
 		},
+		{
+			name:    "gateway openai first output timeout below minimum",
+			mutate:  func(c *Config) { c.Gateway.OpenAIFirstOutputTimeoutSeconds = 29 },
+			wantErr: "gateway.openai_first_output_timeout_seconds",
+		},
+		{
+			name:    "gateway openai high effort first output timeout too large",
+			mutate:  func(c *Config) { c.Gateway.OpenAIHighEffortFirstOutputTimeoutSeconds = 1801 },
+			wantErr: "gateway.openai_high_effort_first_output_timeout_seconds",
+		},
 		{
 			name:    "gateway max idle conns",
 			mutate:  func(c *Config) { c.Gateway.MaxIdleConns = 0 },
@@ -1406,6 +1496,16 @@ func TestValidateConfigErrors(t *testing.T) {
 			mutate:  func(c *Config) { c.Gateway.ImageStreamKeepaliveInterval = -1 },
 			wantErr: "gateway.image_stream_keepalive_interval must be non-negative",
 		},
+		{
+			name:    "gateway image nonstream keepalive range",
+			mutate:  func(c *Config) { c.Gateway.ImageNonstreamKeepaliveInterval = 4 },
+			wantErr: "gateway.image_nonstream_keepalive_interval",
+		},
+		{
+			name:    "gateway image nonstream keepalive negative",
+			mutate:  func(c *Config) { c.Gateway.ImageNonstreamKeepaliveInterval = -1 },
+			wantErr: "gateway.image_nonstream_keepalive_interval must be non-negative",
+		},
 		{
 			name:    "gateway image stream data interval range",
 			mutate:  func(c *Config) { c.Gateway.ImageStreamDataIntervalTimeout = 30 },
@@ -1640,6 +1740,26 @@ func TestValidateConfig_OpenAIWSRules(t *testing.T) {
 			mutate:  func(c *Config) { c.Gateway.OpenAIWS.MaxConnsPerAccount = 0 },
 			wantErr: "gateway.openai_ws.max_conns_per_account",
 		},
+		{
+			name:    "client_first_message_timeout_seconds 必须为正数",
+			mutate:  func(c *Config) { c.Gateway.OpenAIWS.ClientFirstMessageTimeoutSeconds = 0 },
+			wantErr: "gateway.openai_ws.client_first_message_timeout_seconds",
+		},
+		{
+			name:    "client_first_message_timeout_seconds 不能为负数",
+			mutate:  func(c *Config) { c.Gateway.OpenAIWS.ClientFirstMessageTimeoutSeconds = -1 },
+			wantErr: "gateway.openai_ws.client_first_message_timeout_seconds",
+		},
+		{
+			name:    "ingress_inter_turn_idle_timeout_seconds 不能为负数",
+			mutate:  func(c *Config) { c.Gateway.OpenAIWS.IngressInterTurnIdleTimeoutSeconds = -1 },
+			wantErr: "gateway.openai_ws.ingress_inter_turn_idle_timeout_seconds",
+		},
+		{
+			name:    "max_ingress_connections_per_api_key 不能为负数",
+			mutate:  func(c *Config) { c.Gateway.OpenAIWS.MaxIngressConnectionsPerAPIKey = -1 },
+			wantErr: "gateway.openai_ws.max_ingress_connections_per_api_key",
+		},
 		{
 			name:    "min_idle_per_account 不能为负数",
 			mutate:  func(c *Config) { c.Gateway.OpenAIWS.MinIdlePerAccount = -1 },
@@ -1750,6 +1870,42 @@ func TestValidateConfig_OpenAIWSRules(t *testing.T) {
 			mutate:  func(c *Config) { c.Gateway.OpenAIWS.SchedulerScoreWeights.QuotaHeadroom = -0.1 },
 			wantErr: "gateway.openai_ws.scheduler_score_weights.* must be non-negative",
 		},
+		{
+			name:    "scheduler_score_weights upstream_cost 不能为负数",
+			mutate:  func(c *Config) { c.Gateway.OpenAIWS.SchedulerScoreWeights.UpstreamCost = -0.1 },
+			wantErr: "gateway.openai_ws.scheduler_score_weights.* must be non-negative",
+		},
+		{
+			name:    "scheduler_score_weights reset 不能为负数",
+			mutate:  func(c *Config) { c.Gateway.OpenAIWS.SchedulerScoreWeights.Reset = -0.1 },
+			wantErr: "gateway.openai_ws.scheduler_score_weights.* must be non-negative",
+		},
+		{
+			name:    "scheduler_score_weights 不能为 NaN",
+			mutate:  func(c *Config) { c.Gateway.OpenAIWS.SchedulerScoreWeights.PreviousResponse = math.NaN() },
+			wantErr: "gateway.openai_ws.scheduler_score_weights.* must be non-negative and finite",
+		},
+		{
+			name:    "scheduler_score_weights 不能为 Inf",
+			mutate:  func(c *Config) { c.Gateway.OpenAIWS.SchedulerScoreWeights.UpstreamCost = math.Inf(1) },
+			wantErr: "gateway.openai_ws.scheduler_score_weights.* must be non-negative and finite",
+		},
+		{
+			name: "scheduler_score_weights 总和不能溢出",
+			mutate: func(c *Config) {
+				c.Gateway.OpenAIWS.SchedulerScoreWeights.Priority = math.MaxFloat64
+				c.Gateway.OpenAIWS.SchedulerScoreWeights.Load = math.MaxFloat64
+			},
+			wantErr: "gateway.openai_ws.scheduler_score_weights base-weight sum must be finite",
+		},
+		{
+			name: "scheduler_score_weights 含 sticky 总和不能溢出",
+			mutate: func(c *Config) {
+				c.Gateway.OpenAIWS.SchedulerScoreWeights.Priority = math.MaxFloat64
+				c.Gateway.OpenAIWS.SchedulerScoreWeights.PreviousResponse = math.MaxFloat64
+			},
+			wantErr: "gateway.openai_ws.scheduler_score_weights total-weight sum must be finite",
+		},
 		{
 			name: "scheduler_score_weights 不能全为 0",
 			mutate: func(c *Config) {
@@ -1801,6 +1957,30 @@ func TestValidateConfig_OpenAIWSRules(t *testing.T) {
 
 		require.NoError(t, cfg.Validate())
 	})
+
+	t.Run("upstream_cost 可作为唯一有效调度权重", func(t *testing.T) {
+		cfg := buildValid(t)
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.Priority = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.Load = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.Queue = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.ErrorRate = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.TTFT = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.UpstreamCost = 0.1
+
+		require.NoError(t, cfg.Validate())
+	})
+
+	t.Run("reset 可作为唯一有效调度权重", func(t *testing.T) {
+		cfg := buildValid(t)
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.Priority = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.Load = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.Queue = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.ErrorRate = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.TTFT = 0
+		cfg.Gateway.OpenAIWS.SchedulerScoreWeights.Reset = 0.1
+
+		require.NoError(t, cfg.Validate())
+	})
 }
 
 func TestValidateConfig_AutoScaleDisabledIgnoreAutoScaleFields(t *testing.T) {
@@ -1964,6 +2144,9 @@ func TestLoad_DefaultGatewayImageStreamConfig(t *testing.T) {
 	if cfg.Gateway.ImageStreamKeepaliveInterval != 10 {
 		t.Fatalf("image_stream_keepalive_interval = %d, want 10", cfg.Gateway.ImageStreamKeepaliveInterval)
 	}
+	if cfg.Gateway.ImageNonstreamKeepaliveInterval != 0 {
+		t.Fatalf("image_nonstream_keepalive_interval = %d, want 0", cfg.Gateway.ImageNonstreamKeepaliveInterval)
+	}
 	if cfg.Gateway.ImageConcurrency.Enabled {
 		t.Fatalf("image_concurrency.enabled = true, want false")
 	}

backend/internal/handler/admin/account_codex_agent_identity_import_test.go

@@ -0,0 +1,83 @@
+package admin
+
+import (
+	"context"
+	"crypto/ed25519"
+	"crypto/rand"
+	"crypto/x509"
+	"encoding/base64"
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/stretchr/testify/require"
+)
+
+func TestNormalizeCodexImportEntryAcceptsAgentIdentityAuthJSON(t *testing.T) {
+	_, privateKey, err := ed25519.GenerateKey(rand.Reader)
+	require.NoError(t, err)
+	der, err := x509.MarshalPKCS8PrivateKey(privateKey)
+	require.NoError(t, err)
+	privateKeyBase64 := base64.StdEncoding.EncodeToString(der)
+
+	item, err := normalizeCodexImportEntry(codexImportEntry{
+		Index: 1,
+		Value: map[string]any{
+			"auth_mode": "agentIdentity",
+			"agent_identity": map[string]any{
+				"agent_runtime_id":           "runtime-import",
+				"agent_private_key":          privateKeyBase64,
+				"account_id":                 "account-import",
+				"chatgpt_user_id":            "user-import",
+				"email":                      "agent@example.invalid",
+				"plan_type":                  "pro",
+				"chatgpt_account_is_fedramp": false,
+			},
+		},
+	})
+	require.NoError(t, err)
+	require.NotNil(t, item)
+	require.True(t, item.IsAgentIdentity)
+	require.Equal(t, service.OpenAIAuthModeAgentIdentity, item.Credentials["auth_mode"])
+	require.Equal(t, "runtime-import", item.Credentials["agent_runtime_id"])
+	require.Equal(t, privateKeyBase64, item.Credentials["agent_private_key"])
+	require.Equal(t, "account-import", item.Credentials["chatgpt_account_id"])
+	require.Equal(t, "user-import", item.Credentials["chatgpt_user_id"])
+	require.NotContains(t, item.Credentials, "access_token")
+	require.NotContains(t, item.Credentials, "refresh_token")
+	require.NotEmpty(t, item.WarningTexts)
+}
+
+func TestImportCodexSessionsCreatesAgentIdentityWithoutOAuthExpiry(t *testing.T) {
+	_, privateKey, err := ed25519.GenerateKey(rand.Reader)
+	require.NoError(t, err)
+	der, err := x509.MarshalPKCS8PrivateKey(privateKey)
+	require.NoError(t, err)
+
+	svc := newCodexImportMemoryAdminService(nil)
+	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	result, err := handler.importCodexSessions(context.Background(), CodexSessionImportRequest{
+		SkipDefaultGroupBind: boolPtr(true),
+	}, []codexImportEntry{{
+		Index: 1,
+		Value: map[string]any{
+			"auth_mode": "agentIdentity",
+			"agent_identity": map[string]any{
+				"agent_runtime_id":  "runtime-import",
+				"agent_private_key": base64.StdEncoding.EncodeToString(der),
+				"task_id":           "task-import",
+				"account_id":        "account-import",
+				"chatgpt_user_id":   "user-import",
+			},
+		},
+	}})
+	require.NoError(t, err)
+	require.Equal(t, 1, result.Created)
+	require.Zero(t, result.Failed)
+	require.Len(t, svc.createdAccounts, 1)
+	created := svc.createdAccounts[0]
+	require.Nil(t, created.ExpiresAt)
+	require.Nil(t, created.AutoPauseOnExpired)
+	require.Equal(t, service.OpenAIAuthModeAgentIdentity, created.Credentials["auth_mode"])
+	require.NotContains(t, created.Credentials, "access_token")
+	require.NotContains(t, created.Credentials, "refresh_token")
+}

backend/internal/handler/admin/account_codex_import.go

@@ -72,20 +72,25 @@ type codexImportEntry struct {
 }
 
 type codexImportAccount struct {
-	Name           string
-	AccessToken    string
-	RefreshToken   string
-	IDToken        string
-	Email          string
-	AccountID      string
-	UserID         string
-	PlanType       string
-	Organization   string
-	Credentials    map[string]any
-	Extra          map[string]any
-	TokenExpiresAt *time.Time
-	IdentityKeys   []string
-	WarningTexts   []string
+	Name            string
+	AccessToken     string
+	RefreshToken    string
+	IDToken         string
+	Email           string
+	AccountID       string
+	UserID          string
+	PlanType        string
+	Organization    string
+	AgentRuntimeID  string
+	AgentPrivateKey string
+	AgentTaskID     string
+	AgentFedRAMP    bool
+	IsAgentIdentity bool
+	Credentials     map[string]any
+	Extra           map[string]any
+	TokenExpiresAt  *time.Time
+	IdentityKeys    []string
+	WarningTexts    []string
 }
 
 type codexJWTClaims struct {
@@ -115,6 +120,10 @@ func (h *AccountHandler) ImportCodexSession(c *gin.Context) {
 		response.BadRequest(c, "Invalid request: "+err.Error())
 		return
 	}
+	if err := service.ValidateOpenAILongContextBillingExtra(service.PlatformOpenAI, req.Extra); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
 	if req.Concurrency != nil && *req.Concurrency < 0 {
 		response.BadRequest(c, "concurrency must be >= 0")
 		return
@@ -492,6 +501,41 @@ func normalizeCodexImportEntry(entry codexImportEntry) (*codexImportAccount, err
 	case string:
 		item.AccessToken = strings.TrimSpace(raw)
 	case map[string]any:
+		if agentIdentity, ok := firstCodexMap(raw, []string{"agent_identity"}, []string{"agentIdentity"}); ok || strings.EqualFold(firstCodexString(raw, []string{"auth_mode"}, []string{"authMode"}), service.OpenAIAuthModeAgentIdentity) {
+			if !ok {
+				agentIdentity = raw
+			}
+			item.IsAgentIdentity = true
+			item.AgentRuntimeID = firstCodexString(agentIdentity, []string{"agent_runtime_id"}, []string{"agentRuntimeId"})
+			item.AgentPrivateKey = firstCodexString(agentIdentity, []string{"agent_private_key"}, []string{"agentPrivateKey"})
+			item.AgentTaskID = firstCodexString(agentIdentity, []string{"task_id"}, []string{"taskId"})
+			item.AccountID = firstCodexString(agentIdentity, []string{"account_id"}, []string{"accountId"})
+			item.UserID = firstCodexString(agentIdentity, []string{"chatgpt_user_id"}, []string{"chatgptUserId"})
+			item.Email = firstCodexString(agentIdentity, []string{"email"})
+			item.PlanType = firstCodexString(agentIdentity, []string{"plan_type"}, []string{"planType"})
+			item.AgentFedRAMP = firstCodexBool(agentIdentity, []string{"chatgpt_account_is_fedramp"}, []string{"chatgptAccountIsFedramp"})
+			if item.AgentRuntimeID == "" || item.AgentPrivateKey == "" || item.AccountID == "" || item.UserID == "" {
+				return nil, errors.New("agent identity 缺少必要字段")
+			}
+			if err := service.ValidateOpenAIAgentIdentityPrivateKey(item.AgentPrivateKey); err != nil {
+				return nil, errors.New("agent identity private key 格式无效")
+			}
+			item.Credentials["auth_mode"] = service.OpenAIAuthModeAgentIdentity
+			item.Credentials["agent_runtime_id"] = item.AgentRuntimeID
+			item.Credentials["agent_private_key"] = item.AgentPrivateKey
+			item.Credentials["chatgpt_account_id"] = item.AccountID
+			item.Credentials["chatgpt_user_id"] = item.UserID
+			item.Credentials["chatgpt_account_is_fedramp"] = item.AgentFedRAMP
+			setCodexCredentialIfNotEmpty(item.Credentials, "task_id", item.AgentTaskID)
+			setCodexCredentialIfNotEmpty(item.Credentials, "email", item.Email)
+			setCodexCredentialIfNotEmpty(item.Credentials, "plan_type", item.PlanType)
+			if item.AgentTaskID == "" {
+				item.WarningTexts = append(item.WarningTexts, "未包含 task_id，首次请求会使用现有 runtime 注册新 task")
+			}
+			item.IdentityKeys = buildCodexAgentIdentityKeys(item.AccountID, item.UserID, item.Email, item.AgentRuntimeID)
+			item.Name = buildCodexImportAccountName(item, entry.Index)
+			return item, nil
+		}
 		item.AccessToken = firstCodexString(raw,
 			[]string{"tokens", "access_token"},
 			[]string{"tokens", "accessToken"},
@@ -573,6 +617,9 @@ func normalizeCodexImportEntry(entry codexImportEntry) (*codexImportAccount, err
 		return nil, fmt.Errorf("第 %d 条格式不支持", entry.Index)
 	}
 
+	if item.IsAgentIdentity {
+		return item, nil
+	}
 	if item.AccessToken == "" {
 		return nil, errors.New("缺少 accessToken/access_token")
 	}
@@ -727,6 +774,12 @@ func resolveCodexImportExpiry(req CodexSessionImportRequest, item *codexImportAc
 	if item == nil {
 		return nil, nil, nil, nil, errors.New("导入项为空")
 	}
+	// Agent Identity has no OAuth access-token lifetime. Its runtime/task
+	// lifecycle is handled by the upstream task recovery path, so it must not
+	// be rejected or auto-paused by the OAuth import expiry policy.
+	if item.IsAgentIdentity {
+		return nil, nil, nil, nil, nil
+	}
 
 	var requestExpiresAt *time.Time
 	if req.ExpiresAt != nil && *req.ExpiresAt > 0 {
@@ -808,6 +861,9 @@ func sanitizeCodexImportCredentialExtras(input map[string]any) map[string]any {
 		"openai_auth_mode":           {},
 		"token_type":                 {},
 		"chatgpt_account_is_fedramp": {},
+		"agent_runtime_id":           {},
+		"agent_private_key":          {},
+		"task_id":                    {},
 	}
 	out := make(map[string]any, len(input))
 	for key, value := range input {
@@ -838,6 +894,14 @@ func buildCodexImportIdentityKeys(accountID, userID, email, accessToken, refresh
 	return buildCodexStoredIdentityKeys(accountID, userID, email, accessToken)
 }
 
+func buildCodexAgentIdentityKeys(accountID, userID, email, runtimeID string) []string {
+	keys := buildCodexStoredIdentityKeys(accountID, userID, email, "")
+	if runtimeID = strings.TrimSpace(runtimeID); runtimeID != "" {
+		keys = append([]string{"agent:" + runtimeID}, keys...)
+	}
+	return keys
+}
+
 // buildCodexStoredIdentityKeys 生成存量账号索引键，保留 user/account 维度，
 // 让 accessToken-only 账号后续升级为完整 OAuth 时仍能命中并更新原账号。
 func buildCodexStoredIdentityKeys(accountID, userID, email, accessToken string) []string {
@@ -887,6 +951,10 @@ func (i *codexAccountIndex) Add(account service.Account) {
 	for _, key := range keys {
 		i.accountsByKey[key] = upsertCodexAccount(i.accountsByKey[key], account)
 	}
+	if runtimeID := codexCredentialString(account.Credentials, "agent_runtime_id"); runtimeID != "" {
+		key := "agent:" + runtimeID
+		i.accountsByKey[key] = upsertCodexAccount(i.accountsByKey[key], account)
+	}
 }
 
 func (i *codexAccountIndex) remove(accountID int64) {
@@ -1039,6 +1107,38 @@ func firstCodexString(obj map[string]any, paths ...[]string) string {
 	return ""
 }
 
+func firstCodexMap(obj map[string]any, paths ...[]string) (map[string]any, bool) {
+	for _, path := range paths {
+		value, ok := codexPathValue(obj, path)
+		if !ok || value == nil {
+			continue
+		}
+		if mapped, ok := value.(map[string]any); ok {
+			return mapped, true
+		}
+	}
+	return nil, false
+}
+
+func firstCodexBool(obj map[string]any, paths ...[]string) bool {
+	for _, path := range paths {
+		value, ok := codexPathValue(obj, path)
+		if !ok {
+			continue
+		}
+		switch value := value.(type) {
+		case bool:
+			return value
+		case string:
+			parsed, err := strconv.ParseBool(strings.TrimSpace(value))
+			if err == nil {
+				return parsed
+			}
+		}
+	}
+	return false
+}
+
 func copyCodexExtraString(obj map[string]any, extra map[string]any, key string, path []string) {
 	value := firstCodexString(obj, path)
 	if value != "" {

backend/internal/handler/admin/account_codex_import_test.go

@@ -559,7 +559,7 @@ func TestNormalizeCodexImportUsesJWTSubForAccessTokenOnlyIdentity(t *testing.T)
 
 func TestImportCodexSessionsAccessTokenOnlySameWorkspaceDifferentUsersCreatesTwoAccounts(t *testing.T) {
 	svc := newCodexImportMemoryAdminService(nil)
-	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	req := CodexSessionImportRequest{SkipDefaultGroupBind: boolPtr(true)}
 	entries := []codexImportEntry{
 		{Index: 1, Value: buildCodexAccessOnlyImportValue(t, "workspace-1", "user-1")},
@@ -583,7 +583,7 @@ func TestImportCodexSessionsAccessTokenOnlySameWorkspaceDifferentUsersCreatesTwo
 
 func TestImportCodexSessionsAccessTokenOnlySameWorkspaceAndUserDifferentTokensCreatesTwoAccounts(t *testing.T) {
 	svc := newCodexImportMemoryAdminService(nil)
-	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	req := CodexSessionImportRequest{SkipDefaultGroupBind: boolPtr(true)}
 	entries := []codexImportEntry{
 		{Index: 1, Value: map[string]any{
@@ -630,8 +630,9 @@ func TestImportCodexSessionsAccessTokenOnlySameUserUpdatesExisting(t *testing.T)
 			"chatgpt_user_id":    "user-1",
 			"access_token":       existingToken,
 		},
+		Extra: map[string]any{"openai_long_context_billing_enabled": false},
 	}})
-	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	req := CodexSessionImportRequest{SkipDefaultGroupBind: boolPtr(true)}
 	entries := []codexImportEntry{
 		{Index: 1, Value: map[string]any{"access_token": existingToken}},
@@ -650,6 +651,9 @@ func TestImportCodexSessionsAccessTokenOnlySameUserUpdatesExisting(t *testing.T)
 	if len(svc.updatedAccounts) != 1 || svc.updatedAccounts[0].id != 10 {
 		t.Fatalf("updated accounts = %+v, want account 10", svc.updatedAccounts)
 	}
+	if got := svc.updatedAccounts[0].input.Extra["openai_long_context_billing_enabled"]; got != false {
+		t.Fatalf("openai_long_context_billing_enabled = %v, want false", got)
+	}
 }
 
 func TestImportCodexSessionsUpgradesAccessTokenOnlyAccountWithRefreshToken(t *testing.T) {
@@ -666,7 +670,7 @@ func TestImportCodexSessionsUpgradesAccessTokenOnlyAccountWithRefreshToken(t *te
 			"access_token":       oldToken,
 		},
 	}})
-	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	req := CodexSessionImportRequest{SkipDefaultGroupBind: boolPtr(true)}
 	entries := []codexImportEntry{
 		{Index: 1, Value: map[string]any{
@@ -705,7 +709,7 @@ func TestImportCodexSessionsAccessTokenOnlyPreservesExistingRefreshToken(t *test
 			"client_id":          "client-old",
 		},
 	}})
-	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	req := CodexSessionImportRequest{SkipDefaultGroupBind: boolPtr(true)}
 	entries := []codexImportEntry{
 		{Index: 1, Value: map[string]any{"access_token": existingToken}},
@@ -748,7 +752,7 @@ func TestImportCodexSessionsBatchOldAccessTokenDoesNotRollbackRefreshToken(t *te
 			"refresh_token":      "refresh-old",
 		},
 	}})
-	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	req := CodexSessionImportRequest{SkipDefaultGroupBind: boolPtr(true)}
 	entries := []codexImportEntry{
 		{Index: 1, Value: map[string]any{
@@ -794,7 +798,7 @@ func TestImportCodexSessionsWithRefreshTokenKeepsExistingDedup(t *testing.T) {
 			"refresh_token":      "refresh-old",
 		},
 	}})
-	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	req := CodexSessionImportRequest{SkipDefaultGroupBind: boolPtr(true)}
 	entries := []codexImportEntry{
 		{Index: 1, Value: buildCodexRefreshImportValue(t, "workspace-1", "user-1", "refresh-new")},

backend/internal/handler/admin/account_data.go

@@ -460,6 +460,7 @@ func (h *AccountHandler) importData(ctx context.Context, req DataImportRequest)
 		if created.Platform == service.PlatformAntigravity && created.Type == service.AccountTypeOAuth {
 			privacyAccounts = append(privacyAccounts, created)
 		}
+		h.scheduleGrokImportProbe(created)
 		result.AccountCreated++
 	}
 

backend/internal/handler/admin/account_data_handler_test.go

@@ -66,6 +66,7 @@ func setupAccountDataRouter() (*gin.Engine, *stubAdminService) {
 		nil,
 		nil,
 		nil,
+		nil,
 	)
 
 	router.GET("/api/v1/admin/accounts/data", h.ExportData)

backend/internal/handler/admin/account_handler.go

@@ -52,6 +52,7 @@ type AccountHandler struct {
 	openaiOAuthService      *service.OpenAIOAuthService
 	geminiOAuthService      *service.GeminiOAuthService
 	antigravityOAuthService *service.AntigravityOAuthService
+	grokOAuthService        service.GrokOAuthTokenService
 	rateLimitService        *service.RateLimitService
 	accountUsageService     *service.AccountUsageService
 	accountTestService      *service.AccountTestService
@@ -60,6 +61,13 @@ type AccountHandler struct {
 	sessionLimitCache       service.SessionLimitCache
 	rpmCache                service.RPMCache
 	tokenCacheInvalidator   service.TokenCacheInvalidator
+	grokImportProber        grokUsageProber
+	upstreamBillingProbe    *service.UpstreamBillingProbeService
+}
+
+// SetUpstreamBillingProbeService attaches the optional remote billing probe service.
+func (h *AccountHandler) SetUpstreamBillingProbeService(probe *service.UpstreamBillingProbeService) {
+	h.upstreamBillingProbe = probe
 }
 
 // NewAccountHandler creates a new admin account handler
@@ -69,6 +77,7 @@ func NewAccountHandler(
 	openaiOAuthService *service.OpenAIOAuthService,
 	geminiOAuthService *service.GeminiOAuthService,
 	antigravityOAuthService *service.AntigravityOAuthService,
+	grokOAuthService service.GrokOAuthTokenService,
 	rateLimitService *service.RateLimitService,
 	accountUsageService *service.AccountUsageService,
 	accountTestService *service.AccountTestService,
@@ -84,6 +93,7 @@ func NewAccountHandler(
 		openaiOAuthService:      openaiOAuthService,
 		geminiOAuthService:      geminiOAuthService,
 		antigravityOAuthService: antigravityOAuthService,
+		grokOAuthService:        grokOAuthService,
 		rateLimitService:        rateLimitService,
 		accountUsageService:     accountUsageService,
 		accountTestService:      accountTestService,
@@ -784,6 +794,10 @@ func (h *AccountHandler) Create(c *gin.Context) {
 		response.BadRequest(c, "Invalid request: "+err.Error())
 		return
 	}
+	if err := service.ValidateOpenAILongContextBillingExtra(req.Platform, req.Extra); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
 	if req.RateMultiplier != nil && *req.RateMultiplier < 0 {
 		response.BadRequest(c, "rate_multiplier must be >= 0")
 		return
@@ -851,6 +865,54 @@ func (h *AccountHandler) Create(c *gin.Context) {
 	// OpenAI APIKey 账号创建后异步探测上游 /v1/responses 能力。
 	// 探测失败不影响账号创建响应。
 	h.scheduleOpenAIResponsesProbe(createdAccount)
+	h.scheduleGrokImportProbe(createdAccount)
+	response.Success(c, result.Data)
+}
+
+// Duplicate handles creating an independent account from an existing account's configuration.
+// POST /api/v1/admin/accounts/:id/duplicate
+func (h *AccountHandler) Duplicate(c *gin.Context) {
+	accountID, err := strconv.ParseInt(c.Param("id"), 10, 64)
+	if err != nil {
+		response.BadRequest(c, "Invalid account ID")
+		return
+	}
+	actorScope := adminActorScope(c)
+
+	result, err := executeAdminIdempotent(
+		c,
+		"admin.accounts.duplicate",
+		struct {
+			AccountID int64 `json:"account_id"`
+		}{AccountID: accountID},
+		service.DefaultWriteIdempotencyTTL(),
+		func(ctx context.Context) (any, error) {
+			account, execErr := h.adminService.DuplicateAccount(ctx, accountID, actorScope, c.GetHeader("Idempotency-Key"))
+			if execErr != nil {
+				return nil, execErr
+			}
+			return h.buildAccountResponseWithRuntime(ctx, account), nil
+		},
+	)
+	if err != nil {
+		reason := infraerrors.Reason(err)
+		if reason == infraerrors.Reason(service.ErrIdempotencyInProgress) || reason == infraerrors.Reason(service.ErrIdempotencyStoreUnavail) {
+			recovered, recoverErr := h.adminService.RecoverDuplicateAccount(c.Request.Context(), accountID, actorScope, c.GetHeader("Idempotency-Key"))
+			if recoverErr != nil {
+				slog.Warn("account_duplicate_recovery_failed", "account_id", accountID, "actor_scope", actorScope, "reason", reason, "error", recoverErr)
+			} else if recovered != nil {
+				c.Header("X-Idempotency-Recovered", "true")
+				response.Success(c, h.buildAccountResponseWithRuntime(c.Request.Context(), recovered))
+				return
+			}
+		}
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	if result != nil && result.Replayed {
+		c.Header("X-Idempotency-Replayed", "true")
+	}
 	response.Success(c, result.Data)
 }
 
@@ -1173,6 +1235,19 @@ func (h *AccountHandler) refreshSingleAccount(ctx context.Context, account *serv
 				return nil, "", fmt.Errorf("failed to clear account error: %w", clearErr)
 			}
 		}
+	} else if account.Platform == service.PlatformGrok {
+		if h.grokOAuthService == nil {
+			return nil, "", fmt.Errorf("grok oauth service is not configured")
+		}
+		tokenInfo, err := h.grokOAuthService.RefreshAccountToken(ctx, account)
+		if err != nil {
+			return nil, "", fmt.Errorf("failed to refresh Grok credentials: %w", err)
+		}
+
+		newCredentials = service.MergeCredentials(account.Credentials, h.grokOAuthService.BuildAccountCredentials(tokenInfo))
+		if baseURL := strings.TrimSpace(account.GetCredential("base_url")); baseURL != "" {
+			newCredentials["base_url"] = baseURL
+		}
 	} else {
 		// Use Anthropic/Claude OAuth service to refresh token
 		tokenInfo, err := h.oauthService.RefreshAccountToken(ctx, account)
@@ -1299,6 +1374,10 @@ func (h *AccountHandler) ApplyOAuthCredentials(c *gin.Context) {
 		response.ErrorFrom(c, infraerrors.BadRequest("NOT_OAUTH", "cannot apply oauth credentials to non-OAuth account"))
 		return
 	}
+	if err := service.ValidateOpenAILongContextBillingExtra(existing.Platform, req.Extra); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
 
 	updatedAccount, err := h.adminService.UpdateAccount(ctx, accountID, &service.UpdateAccountInput{
 		Type:        req.Type,
@@ -1592,6 +1671,12 @@ func (h *AccountHandler) BatchCreate(c *gin.Context) {
 		response.BadRequest(c, "Invalid request: "+err.Error())
 		return
 	}
+	for _, item := range req.Accounts {
+		if err := service.ValidateOpenAILongContextBillingExtra(item.Platform, item.Extra); err != nil {
+			response.ErrorFrom(c, err)
+			return
+		}
+	}
 
 	executeAdminIdempotentJSON(c, "admin.accounts.batch_create", req, service.DefaultWriteIdempotencyTTL(), func(ctx context.Context) (any, error) {
 		success := 0
@@ -1653,6 +1738,7 @@ func (h *AccountHandler) BatchCreate(c *gin.Context) {
 			}
 			// OpenAI APIKey 账号异步探测 /v1/responses 能力。
 			h.scheduleOpenAIResponsesProbe(account)
+			h.scheduleGrokImportProbe(account)
 			success++
 			results = append(results, gin.H{
 				"name":    item.Name,

backend/internal/handler/admin/account_handler_available_models_test.go

@@ -32,7 +32,7 @@ func (s *availableModelsAdminService) GetAccount(_ context.Context, id int64) (*
 func setupAvailableModelsRouter(adminSvc service.AdminService) *gin.Engine {
 	gin.SetMode(gin.TestMode)
 	router := gin.New()
-	handler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	router.GET("/api/v1/admin/accounts/:id/models", handler.GetAvailableModels)
 	return router
 }
@@ -66,7 +66,7 @@ func setupSyncUpstreamModelsRouter(adminSvc service.AdminService, upstream servi
 		&config.Config{Security: config.SecurityConfig{URLAllowlist: config.URLAllowlistConfig{Enabled: false}}},
 		nil,
 	)
-	handler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, accountTestSvc, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, accountTestSvc, nil, nil, nil, nil, nil)
 	router.POST("/api/v1/admin/accounts/:id/models/sync-upstream", handler.SyncUpstreamModels)
 	return router
 }

backend/internal/handler/admin/account_handler_duplicate_test.go

@@ -0,0 +1,295 @@
+//go:build unit
+
+package admin
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"net/http"
+	"net/http/httptest"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	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"
+)
+
+type duplicateAccountAdminServiceStub struct {
+	service.AdminService
+	account      *service.Account
+	calls        int
+	recoverCalls int
+	accountID    int64
+	actorScope   string
+	operationKey string
+	recoverScope string
+	recoverKey   string
+	recoverErr   error
+	created      bool
+}
+
+type blockingDuplicateAdminServiceStub struct {
+	service.AdminService
+	account      *service.Account
+	started      chan struct{}
+	release      chan struct{}
+	calls        atomic.Int32
+	recoverCalls atomic.Int32
+	recoverErr   error
+}
+
+type failOnceMarkSucceededRepo struct {
+	*memoryIdempotencyRepoStub
+	failNext bool
+}
+
+func (r *failOnceMarkSucceededRepo) MarkSucceeded(ctx context.Context, id int64, responseStatus int, responseBody string, expiresAt time.Time) error {
+	if r.failNext {
+		r.failNext = false
+		return errors.New("mark succeeded failed")
+	}
+	return r.memoryIdempotencyRepoStub.MarkSucceeded(ctx, id, responseStatus, responseBody, expiresAt)
+}
+
+func (s *duplicateAccountAdminServiceStub) DuplicateAccount(_ context.Context, accountID int64, actorScope, operationKey string) (*service.Account, error) {
+	s.calls++
+	s.accountID = accountID
+	s.actorScope = actorScope
+	s.operationKey = operationKey
+	s.created = true
+	return s.account, nil
+}
+
+func (s *duplicateAccountAdminServiceStub) RecoverDuplicateAccount(_ context.Context, _ int64, actorScope, operationKey string) (*service.Account, error) {
+	s.recoverCalls++
+	s.recoverScope = actorScope
+	s.recoverKey = operationKey
+	if s.recoverErr != nil {
+		return nil, s.recoverErr
+	}
+	if !s.created {
+		return nil, nil
+	}
+	return s.account, nil
+}
+
+func (s *blockingDuplicateAdminServiceStub) DuplicateAccount(_ context.Context, _ int64, _, _ string) (*service.Account, error) {
+	s.calls.Add(1)
+	close(s.started)
+	<-s.release
+	return s.account, nil
+}
+
+func (s *blockingDuplicateAdminServiceStub) RecoverDuplicateAccount(_ context.Context, _ int64, _, _ string) (*service.Account, error) {
+	s.recoverCalls.Add(1)
+	return nil, s.recoverErr
+}
+
+func setupDuplicateAccountRouter(t *testing.T, svc service.AdminService) *gin.Engine {
+	t.Helper()
+	previousCoordinator := service.DefaultIdempotencyCoordinator()
+	service.SetDefaultIdempotencyCoordinator(nil)
+	t.Cleanup(func() { service.SetDefaultIdempotencyCoordinator(previousCoordinator) })
+
+	gin.SetMode(gin.TestMode)
+	router := gin.New()
+	router.Use(func(c *gin.Context) {
+		c.Set(string(middleware2.ContextKeyUser), middleware2.AuthSubject{UserID: 77})
+		c.Next()
+	})
+	handler := NewAccountHandler(svc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	router.POST("/api/v1/admin/accounts/:id/duplicate", handler.Duplicate)
+	return router
+}
+
+func TestDuplicateAccountHandlerRedactsCredentials(t *testing.T) {
+	svc := &duplicateAccountAdminServiceStub{
+		account: &service.Account{
+			ID:          43,
+			Name:        "primary (Copy)",
+			Platform:    service.PlatformAnthropic,
+			Type:        service.AccountTypeAPIKey,
+			Status:      service.StatusActive,
+			Schedulable: false,
+			Credentials: map[string]any{"api_key": "top-secret-key"},
+		},
+	}
+	router := setupDuplicateAccountRouter(t, svc)
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(http.MethodPost, "/api/v1/admin/accounts/42/duplicate", nil)
+
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusOK, recorder.Code)
+	require.Equal(t, 1, svc.calls)
+	require.Contains(t, recorder.Body.String(), `"name":"primary (Copy)"`)
+	require.NotContains(t, recorder.Body.String(), "top-secret-key")
+	var responseBody struct {
+		Data struct {
+			Credentials map[string]any `json:"credentials"`
+			Schedulable bool           `json:"schedulable"`
+		} `json:"data"`
+	}
+	require.NoError(t, json.Unmarshal(recorder.Body.Bytes(), &responseBody))
+	require.Empty(t, responseBody.Data.Credentials)
+	require.False(t, responseBody.Data.Schedulable)
+}
+
+func TestDuplicateAccountHandlerRejectsInvalidID(t *testing.T) {
+	svc := &duplicateAccountAdminServiceStub{}
+	router := setupDuplicateAccountRouter(t, svc)
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(http.MethodPost, "/api/v1/admin/accounts/not-a-number/duplicate", nil)
+
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusBadRequest, recorder.Code)
+	require.Zero(t, svc.calls)
+}
+
+func TestDuplicateAccountHandlerReplaysSameIdempotencyKey(t *testing.T) {
+	svc := &duplicateAccountAdminServiceStub{
+		account: &service.Account{
+			ID:          43,
+			Name:        "primary (Copy)",
+			Platform:    service.PlatformAnthropic,
+			Type:        service.AccountTypeAPIKey,
+			Status:      service.StatusActive,
+			Schedulable: false,
+		},
+	}
+	router := setupDuplicateAccountRouter(t, svc)
+	repo := newMemoryIdempotencyRepoStub()
+	service.SetDefaultIdempotencyCoordinator(service.NewIdempotencyCoordinator(repo, service.DefaultIdempotencyConfig()))
+
+	call := func() *httptest.ResponseRecorder {
+		recorder := httptest.NewRecorder()
+		request := httptest.NewRequest(http.MethodPost, "/api/v1/admin/accounts/42/duplicate", nil)
+		request.Header.Set("Idempotency-Key", "duplicate-account-42")
+		router.ServeHTTP(recorder, request)
+		return recorder
+	}
+
+	first := call()
+	second := call()
+
+	require.Equal(t, http.StatusOK, first.Code)
+	require.Equal(t, http.StatusOK, second.Code)
+	require.Equal(t, 1, svc.calls)
+	require.Equal(t, int64(42), svc.accountID)
+	require.Equal(t, "admin:77", svc.actorScope)
+	require.Equal(t, "duplicate-account-42", svc.operationKey)
+	require.Equal(t, "true", second.Header().Get("X-Idempotency-Replayed"))
+}
+
+func TestDuplicateAccountHandlerRecoversAfterMarkSucceededFailure(t *testing.T) {
+	svc := &duplicateAccountAdminServiceStub{
+		account: &service.Account{
+			ID:          43,
+			Name:        "primary (Copy)",
+			Platform:    service.PlatformAnthropic,
+			Type:        service.AccountTypeAPIKey,
+			Status:      service.StatusActive,
+			Schedulable: false,
+		},
+	}
+	router := setupDuplicateAccountRouter(t, svc)
+	repo := &failOnceMarkSucceededRepo{memoryIdempotencyRepoStub: newMemoryIdempotencyRepoStub(), failNext: true}
+	service.SetDefaultIdempotencyCoordinator(service.NewIdempotencyCoordinator(repo, service.DefaultIdempotencyConfig()))
+
+	call := func() *httptest.ResponseRecorder {
+		recorder := httptest.NewRecorder()
+		request := httptest.NewRequest(http.MethodPost, "/api/v1/admin/accounts/42/duplicate", nil)
+		request.Header.Set("Idempotency-Key", "duplicate-account-42-recovery")
+		router.ServeHTTP(recorder, request)
+		return recorder
+	}
+
+	first := call()
+	second := call()
+
+	require.Equal(t, http.StatusOK, first.Code)
+	require.Equal(t, http.StatusOK, second.Code)
+	require.Equal(t, "true", first.Header().Get("X-Idempotency-Recovered"))
+	require.Equal(t, "true", second.Header().Get("X-Idempotency-Recovered"))
+	require.Equal(t, 1, svc.calls, "ambiguous retries must not repeat the create side effect")
+	require.Equal(t, 2, svc.recoverCalls)
+	require.Equal(t, "admin:77", svc.recoverScope)
+	require.Equal(t, "duplicate-account-42-recovery", svc.recoverKey)
+	require.Contains(t, second.Body.String(), `"id":43`)
+}
+
+func TestDuplicateAccountHandlerPreservesIdempotencyErrorWhenRecoveryLookupFails(t *testing.T) {
+	svc := &duplicateAccountAdminServiceStub{
+		account: &service.Account{
+			ID:          43,
+			Name:        "primary (Copy)",
+			Platform:    service.PlatformAnthropic,
+			Type:        service.AccountTypeAPIKey,
+			Status:      service.StatusActive,
+			Schedulable: false,
+		},
+		recoverErr: errors.New("recovery database unavailable"),
+	}
+	router := setupDuplicateAccountRouter(t, svc)
+	repo := &failOnceMarkSucceededRepo{memoryIdempotencyRepoStub: newMemoryIdempotencyRepoStub(), failNext: true}
+	service.SetDefaultIdempotencyCoordinator(service.NewIdempotencyCoordinator(repo, service.DefaultIdempotencyConfig()))
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(http.MethodPost, "/api/v1/admin/accounts/42/duplicate", nil)
+	request.Header.Set("Idempotency-Key", "duplicate-account-42-recovery-error")
+
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusServiceUnavailable, recorder.Code)
+	require.Contains(t, recorder.Body.String(), "IDEMPOTENCY_STORE_UNAVAILABLE")
+	require.Equal(t, 1, svc.calls)
+	require.Equal(t, 1, svc.recoverCalls)
+	require.Equal(t, "admin:77", svc.recoverScope)
+}
+
+func TestDuplicateAccountHandlerDoesNotReexecuteWhileOriginalIsProcessing(t *testing.T) {
+	svc := &blockingDuplicateAdminServiceStub{
+		account: &service.Account{
+			ID:          43,
+			Name:        "primary (Copy)",
+			Platform:    service.PlatformAnthropic,
+			Type:        service.AccountTypeAPIKey,
+			Status:      service.StatusActive,
+			Schedulable: false,
+		},
+		started:    make(chan struct{}),
+		release:    make(chan struct{}),
+		recoverErr: errors.New("recovery database unavailable"),
+	}
+	router := setupDuplicateAccountRouter(t, svc)
+	service.SetDefaultIdempotencyCoordinator(service.NewIdempotencyCoordinator(newMemoryIdempotencyRepoStub(), service.DefaultIdempotencyConfig()))
+
+	call := func() *httptest.ResponseRecorder {
+		recorder := httptest.NewRecorder()
+		request := httptest.NewRequest(http.MethodPost, "/api/v1/admin/accounts/42/duplicate", nil)
+		request.Header.Set("Idempotency-Key", "duplicate-account-42-active")
+		router.ServeHTTP(recorder, request)
+		return recorder
+	}
+	firstDone := make(chan *httptest.ResponseRecorder, 1)
+	go func() { firstDone <- call() }()
+	<-svc.started
+
+	second := call()
+	require.Equal(t, http.StatusConflict, second.Code)
+	require.Contains(t, second.Body.String(), "IDEMPOTENCY_IN_PROGRESS")
+	require.Equal(t, int32(1), svc.calls.Load())
+	require.Equal(t, int32(1), svc.recoverCalls.Load())
+
+	close(svc.release)
+	select {
+	case first := <-firstDone:
+		require.Equal(t, http.StatusOK, first.Code)
+	case <-time.After(time.Second):
+		t.Fatal("original duplicate request did not finish")
+	}
+}

backend/internal/handler/admin/account_handler_grok_refresh_test.go

@@ -0,0 +1,99 @@
+//go:build unit
+
+package admin
+
+import (
+	"context"
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/stretchr/testify/require"
+)
+
+type grokRefreshOAuthStub struct {
+	account *service.Account
+	info    *service.GrokTokenInfo
+	calls   int
+}
+
+func (s *grokRefreshOAuthStub) RefreshAccountToken(_ context.Context, account *service.Account) (*service.GrokTokenInfo, error) {
+	s.calls++
+	s.account = account
+	return s.info, nil
+}
+
+func (s *grokRefreshOAuthStub) BuildAccountCredentials(info *service.GrokTokenInfo) map[string]any {
+	return map[string]any{
+		"access_token":  info.AccessToken,
+		"refresh_token": info.RefreshToken,
+		"expires_at":    info.ExpiresAt,
+		"base_url":      "https://api.x.ai/v1",
+	}
+}
+
+type grokRefreshAdminService struct {
+	*stubAdminService
+	updatedCredentials map[string]any
+}
+
+func (s *grokRefreshAdminService) UpdateAccount(_ context.Context, id int64, input *service.UpdateAccountInput) (*service.Account, error) {
+	s.updatedCredentials = input.Credentials
+	return &service.Account{
+		ID:          id,
+		Platform:    service.PlatformGrok,
+		Type:        service.AccountTypeOAuth,
+		Status:      service.StatusActive,
+		Credentials: input.Credentials,
+	}, nil
+}
+
+func TestRefreshSingleAccountRoutesGrokThroughGrokOAuthService(t *testing.T) {
+	t.Parallel()
+
+	adminSvc := &grokRefreshAdminService{stubAdminService: newStubAdminService()}
+	grokOAuth := &grokRefreshOAuthStub{info: &service.GrokTokenInfo{
+		AccessToken:  "new-access",
+		RefreshToken: "new-refresh",
+		ExpiresAt:    1_800_000_000,
+	}}
+	handler := NewAccountHandler(
+		adminSvc,
+		nil,
+		nil,
+		nil,
+		nil,
+		grokOAuth,
+		nil,
+		nil,
+		nil,
+		nil,
+		nil,
+		nil,
+		nil,
+		nil,
+	)
+	account := &service.Account{
+		ID:       4227,
+		Platform: service.PlatformGrok,
+		Type:     service.AccountTypeOAuth,
+		Credentials: map[string]any{
+			"access_token":       "old-access",
+			"refresh_token":      "old-refresh",
+			"base_url":           "https://example.invalid/v1",
+			"subscription_tier":  "SUPER_GROK",
+			"entitlement_status": "ACTIVE",
+		},
+	}
+
+	updated, warning, err := handler.refreshSingleAccount(context.Background(), account)
+	require.NoError(t, err)
+	require.Empty(t, warning)
+	require.Equal(t, 1, grokOAuth.calls)
+	require.Same(t, account, grokOAuth.account)
+	require.Equal(t, "new-access", adminSvc.updatedCredentials["access_token"])
+	require.Equal(t, "new-refresh", adminSvc.updatedCredentials["refresh_token"])
+	require.Equal(t, "https://example.invalid/v1", adminSvc.updatedCredentials["base_url"])
+	require.Equal(t, "SUPER_GROK", adminSvc.updatedCredentials["subscription_tier"])
+	require.Equal(t, "ACTIVE", adminSvc.updatedCredentials["entitlement_status"])
+	require.Equal(t, adminSvc.updatedCredentials, updated.Credentials)
+}

backend/internal/handler/admin/account_handler_list_test.go

@@ -17,7 +17,7 @@ func setupAccountListRouter() (*gin.Engine, *stubAdminService) {
 	gin.SetMode(gin.TestMode)
 	router := gin.New()
 	adminSvc := newStubAdminService()
-	handler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	router.GET("/api/v1/admin/accounts", handler.List)
 	return router, adminSvc
 }

backend/internal/handler/admin/account_handler_long_context_billing_test.go

@@ -0,0 +1,165 @@
+package admin
+
+import (
+	"bytes"
+	"encoding/json"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	infraerrors "github.com/Wei-Shaw/sub2api/internal/pkg/errors"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/require"
+)
+
+func TestAccountAdminBoundariesRejectMalformedOpenAILongContextBillingValue(t *testing.T) {
+	const malformedExtra = `"extra":{"openai_long_context_billing_enabled":"true"}`
+
+	tests := []struct {
+		name   string
+		method string
+		path   string
+		body   string
+		mount  func(*gin.Engine, *AccountHandler)
+		setup  func(*stubAdminService)
+	}{
+		{
+			name:   "create",
+			method: http.MethodPost,
+			path:   "/accounts",
+			body:   `{"name":"account","platform":"openai","type":"apikey","credentials":{"api_key":"test"},` + malformedExtra + `}`,
+			mount:  func(router *gin.Engine, handler *AccountHandler) { router.POST("/accounts", handler.Create) },
+		},
+		{
+			name:   "update",
+			method: http.MethodPut,
+			path:   "/accounts/1",
+			body:   `{` + malformedExtra + `}`,
+			mount:  func(router *gin.Engine, handler *AccountHandler) { router.PUT("/accounts/:id", handler.Update) },
+			setup: func(stub *stubAdminService) {
+				stub.updateAccountErr = infraerrors.BadRequest("OPENAI_LONG_CONTEXT_BILLING_INVALID", "invalid")
+			},
+		},
+		{
+			name:   "bulk update",
+			method: http.MethodPost,
+			path:   "/accounts/bulk-update",
+			body:   `{"account_ids":[1],` + malformedExtra + `}`,
+			mount: func(router *gin.Engine, handler *AccountHandler) {
+				router.POST("/accounts/bulk-update", handler.BulkUpdate)
+			},
+			setup: func(stub *stubAdminService) {
+				stub.bulkUpdateAccountErr = infraerrors.BadRequest("OPENAI_LONG_CONTEXT_BILLING_INVALID", "invalid")
+			},
+		},
+		{
+			name:   "batch create",
+			method: http.MethodPost,
+			path:   "/accounts/batch",
+			body:   `{"accounts":[{"name":"account","platform":"openai","type":"apikey","credentials":{"api_key":"test"},` + malformedExtra + `}]}`,
+			mount:  func(router *gin.Engine, handler *AccountHandler) { router.POST("/accounts/batch", handler.BatchCreate) },
+		},
+		{
+			name:   "Codex session import",
+			method: http.MethodPost,
+			path:   "/accounts/import-codex-session",
+			body:   `{"content":"token",` + malformedExtra + `}`,
+			mount: func(router *gin.Engine, handler *AccountHandler) {
+				router.POST("/accounts/import-codex-session", handler.ImportCodexSession)
+			},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			gin.SetMode(gin.TestMode)
+			stub := newStubAdminService()
+			if tt.setup != nil {
+				tt.setup(stub)
+			}
+			handler := NewAccountHandler(stub, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+			router := gin.New()
+			tt.mount(router, handler)
+			recorder := httptest.NewRecorder()
+			request := httptest.NewRequest(tt.method, tt.path, bytes.NewBufferString(tt.body))
+			request.Header.Set("Content-Type", "application/json")
+
+			router.ServeHTTP(recorder, request)
+
+			require.Equal(t, http.StatusBadRequest, recorder.Code)
+			var responseBody struct {
+				Reason string `json:"reason"`
+			}
+			require.NoError(t, json.Unmarshal(recorder.Body.Bytes(), &responseBody))
+			require.Equal(t, "OPENAI_LONG_CONTEXT_BILLING_INVALID", responseBody.Reason)
+		})
+	}
+}
+
+func TestAccountCreateBoundaryDoesNotApplyOpenAIValidationToOtherPlatforms(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	handler := NewAccountHandler(newStubAdminService(), nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	router := gin.New()
+	router.POST("/accounts", handler.Create)
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(http.MethodPost, "/accounts", bytes.NewBufferString(
+		`{"name":"account","platform":"anthropic","type":"apikey","credentials":{"api_key":"test"},"extra":{"openai_long_context_billing_enabled":"provider-owned"}}`,
+	))
+	request.Header.Set("Content-Type", "application/json")
+
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusOK, recorder.Code)
+}
+
+func TestApplyOAuthCredentialsRejectsMalformedOpenAILongContextBillingBeforeMutation(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	stub := newStubAdminService()
+	stub.getAccountResult = &service.Account{
+		ID:       1,
+		Platform: service.PlatformOpenAI,
+		Type:     service.AccountTypeOAuth,
+	}
+	handler := NewAccountHandler(stub, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	router := gin.New()
+	router.POST("/accounts/:id/apply-oauth-credentials", handler.ApplyOAuthCredentials)
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(http.MethodPost, "/accounts/1/apply-oauth-credentials", bytes.NewBufferString(
+		`{"type":"oauth","credentials":{"access_token":"new-token"},"extra":{"openai_long_context_billing_enabled":"true"}}`,
+	))
+	request.Header.Set("Content-Type", "application/json")
+
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusBadRequest, recorder.Code)
+	var responseBody struct {
+		Reason string `json:"reason"`
+	}
+	require.NoError(t, json.Unmarshal(recorder.Body.Bytes(), &responseBody))
+	require.Equal(t, "OPENAI_LONG_CONTEXT_BILLING_INVALID", responseBody.Reason)
+	require.Zero(t, stub.updateAccountCalls)
+	require.Zero(t, stub.updateAccountExtraCalls)
+}
+
+func TestOpenAIOAuthCodexPATBoundaryRejectsMalformedOpenAILongContextBillingValueBeforeTokenValidation(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	handler := NewOpenAIOAuthHandler(nil, newStubAdminService(), nil)
+	router := gin.New()
+	router.Use(gin.Recovery())
+	router.POST("/openai/create-from-codex-pat", handler.CreateAccountFromCodexPAT)
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(http.MethodPost, "/openai/create-from-codex-pat", bytes.NewBufferString(
+		`{"access_token":"token","extra":{"openai_long_context_billing_enabled":1}}`,
+	))
+	request.Header.Set("Content-Type", "application/json")
+
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusBadRequest, recorder.Code)
+	var responseBody struct {
+		Reason string `json:"reason"`
+	}
+	require.NoError(t, json.Unmarshal(recorder.Body.Bytes(), &responseBody))
+	require.Equal(t, "OPENAI_LONG_CONTEXT_BILLING_INVALID", responseBody.Reason)
+}

backend/internal/handler/admin/account_handler_mixed_channel_test.go

@@ -15,7 +15,7 @@ import (
 func setupAccountMixedChannelRouter(adminSvc *stubAdminService) *gin.Engine {
 	gin.SetMode(gin.TestMode)
 	router := gin.New()
-	accountHandler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	accountHandler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	router.POST("/api/v1/admin/accounts/check-mixed-channel", accountHandler.CheckMixedChannel)
 	router.POST("/api/v1/admin/accounts", accountHandler.Create)
 	router.PUT("/api/v1/admin/accounts/:id", accountHandler.Update)

backend/internal/handler/admin/account_handler_passthrough_test.go

@@ -29,6 +29,7 @@ func TestAccountHandler_Create_AnthropicAPIKeyPassthroughExtraForwarded(t *testi
 		nil,
 		nil,
 		nil,
+		nil,
 	)
 
 	router := gin.New()

backend/internal/handler/admin/account_upstream_billing_probe.go

@@ -0,0 +1,122 @@
+package admin
+
+import (
+	"strconv"
+
+	"github.com/Wei-Shaw/sub2api/internal/pkg/response"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/gin-gonic/gin"
+)
+
+type upstreamBillingProbeEnabledRequest struct {
+	Enabled *bool `json:"enabled" binding:"required"`
+}
+
+type upstreamBillingProbeBatchRequest struct {
+	AccountIDs []int64 `json:"account_ids" binding:"required"`
+}
+
+func (h *AccountHandler) GetUpstreamBillingProbeSettings(c *gin.Context) {
+	if h.upstreamBillingProbe == nil {
+		response.ErrorFrom(c, service.ErrUpstreamBillingProbeUnavailable)
+		return
+	}
+	settings, err := h.upstreamBillingProbe.GetSettings(c.Request.Context())
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, settings)
+}
+
+func (h *AccountHandler) UpdateUpstreamBillingProbeSettings(c *gin.Context) {
+	if h.upstreamBillingProbe == nil {
+		response.ErrorFrom(c, service.ErrUpstreamBillingProbeUnavailable)
+		return
+	}
+	var req service.UpstreamBillingProbeSettings
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.BadRequest(c, "Invalid request: "+err.Error())
+		return
+	}
+	if err := h.upstreamBillingProbe.UpdateSettings(c.Request.Context(), &req); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	settings, err := h.upstreamBillingProbe.GetSettings(c.Request.Context())
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, settings)
+}
+
+func (h *AccountHandler) SetUpstreamBillingProbeEnabled(c *gin.Context) {
+	if h.upstreamBillingProbe == nil {
+		response.ErrorFrom(c, service.ErrUpstreamBillingProbeUnavailable)
+		return
+	}
+	accountID, err := strconv.ParseInt(c.Param("id"), 10, 64)
+	if err != nil || accountID <= 0 {
+		response.BadRequest(c, "Invalid account ID")
+		return
+	}
+	var req upstreamBillingProbeEnabledRequest
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.BadRequest(c, "Invalid request: "+err.Error())
+		return
+	}
+	if err := h.upstreamBillingProbe.SetAccountEnabled(c.Request.Context(), accountID, *req.Enabled); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, gin.H{"account_id": accountID, "enabled": *req.Enabled})
+}
+
+func (h *AccountHandler) ProbeUpstreamBilling(c *gin.Context) {
+	if h.upstreamBillingProbe == nil {
+		response.ErrorFrom(c, service.ErrUpstreamBillingProbeUnavailable)
+		return
+	}
+	accountID, err := strconv.ParseInt(c.Param("id"), 10, 64)
+	if err != nil || accountID <= 0 {
+		response.BadRequest(c, "Invalid account ID")
+		return
+	}
+	snapshot, err := h.upstreamBillingProbe.ProbeAccount(c.Request.Context(), accountID)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, service.UpstreamBillingProbeResult{AccountID: accountID, Snapshot: snapshot})
+}
+
+func (h *AccountHandler) ProbeUpstreamBillingBatch(c *gin.Context) {
+	if h.upstreamBillingProbe == nil {
+		response.ErrorFrom(c, service.ErrUpstreamBillingProbeUnavailable)
+		return
+	}
+	var req upstreamBillingProbeBatchRequest
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.BadRequest(c, "Invalid request: "+err.Error())
+		return
+	}
+	if len(req.AccountIDs) == 0 || len(req.AccountIDs) > service.UpstreamBillingProbeMaxBatchSize {
+		response.BadRequest(c, "account_ids must contain between 1 and 20 items")
+		return
+	}
+	seen := make(map[int64]struct{}, len(req.AccountIDs))
+	accountIDs := make([]int64, 0, len(req.AccountIDs))
+	for _, accountID := range req.AccountIDs {
+		if accountID <= 0 {
+			response.BadRequest(c, "account_ids must contain positive IDs")
+			return
+		}
+		if _, exists := seen[accountID]; exists {
+			continue
+		}
+		seen[accountID] = struct{}{}
+		accountIDs = append(accountIDs, accountID)
+	}
+	response.Success(c, gin.H{"results": h.upstreamBillingProbe.ProbeAccounts(c.Request.Context(), accountIDs)})
+}

backend/internal/handler/admin/account_upstream_billing_probe_test.go

@@ -0,0 +1,71 @@
+package admin
+
+import (
+	"bytes"
+	"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 setupUpstreamBillingProbeRouter() *gin.Engine {
+	gin.SetMode(gin.TestMode)
+	handler := NewAccountHandler(nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler.SetUpstreamBillingProbeService(service.NewUpstreamBillingProbeService(nil, nil, nil))
+
+	router := gin.New()
+	router.GET("/admin/accounts/upstream-billing-probe/settings", handler.GetUpstreamBillingProbeSettings)
+	router.POST("/admin/accounts/upstream-billing-probe/batch", handler.ProbeUpstreamBillingBatch)
+	router.PUT("/admin/accounts/:id/upstream-billing-probe", handler.SetUpstreamBillingProbeEnabled)
+	return router
+}
+
+func TestAccountHandlerGetUpstreamBillingProbeSettingsReturnsDefaults(t *testing.T) {
+	router := setupUpstreamBillingProbeRouter()
+	recorder := httptest.NewRecorder()
+	router.ServeHTTP(recorder, httptest.NewRequest(http.MethodGet, "/admin/accounts/upstream-billing-probe/settings", nil))
+
+	require.Equal(t, http.StatusOK, recorder.Code)
+	var response struct {
+		Data service.UpstreamBillingProbeSettings `json:"data"`
+	}
+	require.NoError(t, json.Unmarshal(recorder.Body.Bytes(), &response))
+	require.True(t, response.Data.Enabled)
+	require.Equal(t, 30, response.Data.IntervalMinutes)
+}
+
+func TestAccountHandlerProbeUpstreamBillingBatchValidatesIDs(t *testing.T) {
+	router := setupUpstreamBillingProbeRouter()
+
+	for _, body := range []string{`{"account_ids":[]}`, `{"account_ids":[0]}`} {
+		recorder := httptest.NewRecorder()
+		request := httptest.NewRequest(http.MethodPost, "/admin/accounts/upstream-billing-probe/batch", bytes.NewBufferString(body))
+		request.Header.Set("Content-Type", "application/json")
+		router.ServeHTTP(recorder, request)
+		require.Equal(t, http.StatusBadRequest, recorder.Code)
+	}
+}
+
+func TestAccountHandlerSetUpstreamBillingProbeEnabledRejectsInvalidID(t *testing.T) {
+	router := setupUpstreamBillingProbeRouter()
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(http.MethodPut, "/admin/accounts/not-an-id/upstream-billing-probe", bytes.NewBufferString(`{"enabled":true}`))
+	request.Header.Set("Content-Type", "application/json")
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusBadRequest, recorder.Code)
+}
+
+func TestAccountHandlerSetUpstreamBillingProbeEnabledRequiresValue(t *testing.T) {
+	router := setupUpstreamBillingProbeRouter()
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(http.MethodPut, "/admin/accounts/1/upstream-billing-probe", bytes.NewBufferString(`{}`))
+	request.Header.Set("Content-Type", "application/json")
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusBadRequest, recorder.Code)
+}

backend/internal/handler/admin/admin_basic_handlers_test.go

@@ -16,7 +16,7 @@ func setupAdminRouter() (*gin.Engine, *stubAdminService) {
 	router := gin.New()
 	adminSvc := newStubAdminService()
 
-	userHandler := NewUserHandler(adminSvc, nil, nil, nil)
+	userHandler := NewUserHandler(adminSvc, nil, nil, nil, nil, nil)
 	groupHandler := NewGroupHandler(adminSvc, nil, nil)
 	proxyHandler := NewProxyHandler(adminSvc)
 	redeemHandler := NewRedeemHandler(adminSvc, nil)

backend/internal/handler/admin/admin_service_stub_test.go

@@ -33,6 +33,9 @@ type stubAdminService struct {
 	createSparkShadowErr                error
 	updateAccountErr                    error
 	bulkUpdateAccountErr                error
+	getAccountResult                    *service.Account
+	updateAccountCalls                  int
+	updateAccountExtraCalls             int
 	checkMixedErr                       error
 	lastMixedCheck                      struct {
 		accountID int64
@@ -388,6 +391,9 @@ func (s *stubAdminService) ListOpenAISchedulableAccountsForSchedulerScore(_ cont
 }
 
 func (s *stubAdminService) GetAccount(ctx context.Context, id int64) (*service.Account, error) {
+	if s.getAccountResult != nil {
+		return s.getAccountResult, nil
+	}
 	account := service.Account{ID: id, Name: "account", Status: service.StatusActive}
 	return &account, nil
 }
@@ -412,7 +418,17 @@ func (s *stubAdminService) CreateAccount(ctx context.Context, input *service.Cre
 	return &account, nil
 }
 
+func (s *stubAdminService) DuplicateAccount(ctx context.Context, id int64, actorScope, operationKey string) (*service.Account, error) {
+	account := service.Account{ID: 301, Name: "account (Copy)", Status: service.StatusActive, Schedulable: false}
+	return &account, nil
+}
+
+func (s *stubAdminService) RecoverDuplicateAccount(ctx context.Context, id int64, actorScope, operationKey string) (*service.Account, error) {
+	return nil, nil
+}
+
 func (s *stubAdminService) UpdateAccount(ctx context.Context, id int64, input *service.UpdateAccountInput) (*service.Account, error) {
+	s.updateAccountCalls++
 	if s.updateAccountErr != nil {
 		return nil, s.updateAccountErr
 	}
@@ -421,6 +437,7 @@ func (s *stubAdminService) UpdateAccount(ctx context.Context, id int64, input *s
 }
 
 func (s *stubAdminService) UpdateAccountExtra(ctx context.Context, id int64, updates map[string]any) error {
+	s.updateAccountExtraCalls++
 	return nil
 }
 

backend/internal/handler/admin/audit_log_handler.go

@@ -0,0 +1,170 @@
+package admin
+
+import (
+	"net/http"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/Wei-Shaw/sub2api/internal/pkg/ip"
+	"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"
+)
+
+// AuditLogHandler 操作审计日志管理接口。
+// 审计日志仅管理员可见；不提供单条删除，仅支持带 TOTP 验证的全量清空。
+type AuditLogHandler struct {
+	auditService *service.AuditLogService
+	totpService  *service.TotpService
+}
+
+// NewAuditLogHandler 创建审计日志处理器。
+func NewAuditLogHandler(auditService *service.AuditLogService, totpService *service.TotpService) *AuditLogHandler {
+	return &AuditLogHandler{
+		auditService: auditService,
+		totpService:  totpService,
+	}
+}
+
+// List 分页查询审计日志。
+// GET /api/v1/admin/audit-logs
+func (h *AuditLogHandler) List(c *gin.Context) {
+	page, pageSize := response.ParsePagination(c)
+	if pageSize > 200 {
+		pageSize = 200
+	}
+
+	filter := &service.AuditLogFilter{
+		Page:       page,
+		PageSize:   pageSize,
+		ActorEmail: strings.TrimSpace(c.Query("actor_email")),
+		AuthMethod: strings.TrimSpace(c.Query("auth_method")),
+		Action:     strings.TrimSpace(c.Query("action")),
+		Method:     strings.TrimSpace(c.Query("method")),
+		ClientIP:   strings.TrimSpace(c.Query("client_ip")),
+		Query:      strings.TrimSpace(c.Query("q")),
+	}
+
+	if v := strings.TrimSpace(c.Query("actor_user_id")); v != "" {
+		id, err := strconv.ParseInt(v, 10, 64)
+		if err != nil || id <= 0 {
+			response.BadRequest(c, "Invalid actor_user_id")
+			return
+		}
+		filter.ActorUserID = &id
+	}
+	if v := strings.TrimSpace(c.Query("start_time")); v != "" {
+		t, err := time.Parse(time.RFC3339, v)
+		if err != nil {
+			response.BadRequest(c, "Invalid start_time, expect RFC3339")
+			return
+		}
+		filter.StartTime = &t
+	}
+	if v := strings.TrimSpace(c.Query("end_time")); v != "" {
+		t, err := time.Parse(time.RFC3339, v)
+		if err != nil {
+			response.BadRequest(c, "Invalid end_time, expect RFC3339")
+			return
+		}
+		filter.EndTime = &t
+	}
+	if v := strings.TrimSpace(c.Query("success")); v != "" {
+		success := v == "true"
+		filter.Success = &success
+	}
+
+	result, err := h.auditService.List(c.Request.Context(), filter)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Paginated(c, result.Logs, int64(result.Total), result.Page, result.PageSize)
+}
+
+// Get 查询单条审计日志详情（含脱敏后的请求体）。
+// GET /api/v1/admin/audit-logs/:id
+func (h *AuditLogHandler) Get(c *gin.Context) {
+	id, err := strconv.ParseInt(strings.TrimSpace(c.Param("id")), 10, 64)
+	if err != nil || id <= 0 {
+		response.BadRequest(c, "Invalid audit log id")
+		return
+	}
+	item, err := h.auditService.GetByID(c.Request.Context(), id)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, item)
+}
+
+type auditLogClearRequest struct {
+	TotpCode string `json:"totp_code" binding:"required"`
+}
+
+// Clear 全量清空审计日志。
+// POST /api/v1/admin/audit-logs/clear
+//
+// 安全要求（与需求对齐）：
+//  1. 每次清空都必须现场验证 TOTP 码（不复用 step-up sudo 窗口）
+//  2. 未启用 TOTP 的管理员不允许清空
+//  3. admin API key（机器凭证）不允许清空
+//  4. 清空完成后同步写入一条留痕记录（操作者、IP、UA、删除行数）
+func (h *AuditLogHandler) Clear(c *gin.Context) {
+	if c.GetString("auth_method") == service.AuditAuthMethodAdminAPIKey {
+		response.ErrorWithDetails(c, http.StatusForbidden,
+			"Admin API key cannot clear audit logs; a two-factor verified admin session is required",
+			"STEP_UP_ADMIN_API_KEY_FORBIDDEN", nil)
+		return
+	}
+
+	subject, ok := middleware.GetAuthSubjectFromContext(c)
+	if !ok || subject.UserID <= 0 {
+		response.Unauthorized(c, "Unauthorized")
+		return
+	}
+
+	var req auditLogClearRequest
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.ErrorWithDetails(c, http.StatusBadRequest,
+			"TOTP code is required to clear audit logs", "TOTP_CODE_REQUIRED", nil)
+		return
+	}
+
+	// 校验 TOTP：未启用（ErrTotpNotSetup）、码错误、尝试过多均在此拦截。
+	if err := h.totpService.VerifyCode(c.Request.Context(), subject.UserID, strings.TrimSpace(req.TotpCode)); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	uid := subject.UserID
+	role, _ := middleware.GetUserRoleFromContext(c)
+	trace := &service.AuditLog{
+		ActorUserID:      &uid,
+		ActorEmail:       c.GetString(middleware.ContextKeyAuthEmail),
+		ActorRole:        role,
+		AuthMethod:       c.GetString("auth_method"),
+		CredentialMasked: middleware.MaskedRequestCredential(c),
+		Method:           http.MethodPost,
+		Path:             c.FullPath(),
+		ClientIP:         ip.GetTrustedClientIP(c),
+		UserAgent:        c.Request.UserAgent(),
+		StatusCode:       http.StatusOK,
+	}
+	if requestID := c.Writer.Header().Get("X-Request-ID"); requestID != "" {
+		trace.RequestID = requestID
+	}
+
+	deleted, err := h.auditService.ClearAll(c.Request.Context(), trace)
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	// 留痕记录已同步落库，跳过异步审计中间件的重复记录。
+	middleware.SkipAudit(c)
+	response.Success(c, gin.H{"deleted": deleted})
+}

backend/internal/handler/admin/batch_update_credentials_test.go

@@ -36,7 +36,7 @@ func (f *failingAdminService) UpdateAccount(ctx context.Context, id int64, input
 func setupAccountHandlerWithService(adminSvc service.AdminService) (*gin.Engine, *AccountHandler) {
 	gin.SetMode(gin.TestMode)
 	router := gin.New()
-	handler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	handler := NewAccountHandler(adminSvc, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	router.POST("/api/v1/admin/accounts/batch-update-credentials", handler.BatchUpdateCredentials)
 	return router, handler
 }

backend/internal/handler/admin/channel_handler.go

@@ -64,6 +64,7 @@ type channelModelPricingRequest struct {
 	OutputPrice      *float64                 `json:"output_price" binding:"omitempty,min=0"`
 	CacheWritePrice  *float64                 `json:"cache_write_price" binding:"omitempty,min=0"`
 	CacheReadPrice   *float64                 `json:"cache_read_price" binding:"omitempty,min=0"`
+	ImageInputPrice  *float64                 `json:"image_input_price" binding:"omitempty,min=0"`
 	ImageOutputPrice *float64                 `json:"image_output_price" binding:"omitempty,min=0"`
 	PerRequestPrice  *float64                 `json:"per_request_price" binding:"omitempty,min=0"`
 	Intervals        []pricingIntervalRequest `json:"intervals"`
@@ -115,6 +116,7 @@ type channelModelPricingResponse struct {
 	OutputPrice      *float64                  `json:"output_price"`
 	CacheWritePrice  *float64                  `json:"cache_write_price"`
 	CacheReadPrice   *float64                  `json:"cache_read_price"`
+	ImageInputPrice  *float64                  `json:"image_input_price"`
 	ImageOutputPrice *float64                  `json:"image_output_price"`
 	PerRequestPrice  *float64                  `json:"per_request_price"`
 	Intervals        []pricingIntervalResponse `json:"intervals"`
@@ -222,6 +224,7 @@ func pricingToResponse(p *service.ChannelModelPricing) channelModelPricingRespon
 		OutputPrice:      p.OutputPrice,
 		CacheWritePrice:  p.CacheWritePrice,
 		CacheReadPrice:   p.CacheReadPrice,
+		ImageInputPrice:  p.ImageInputPrice,
 		ImageOutputPrice: p.ImageOutputPrice,
 		PerRequestPrice:  p.PerRequestPrice,
 		Intervals:        intervals,
@@ -273,6 +276,7 @@ func pricingRequestToService(reqs []channelModelPricingRequest) []service.Channe
 			OutputPrice:      r.OutputPrice,
 			CacheWritePrice:  r.CacheWritePrice,
 			CacheReadPrice:   r.CacheReadPrice,
+			ImageInputPrice:  r.ImageInputPrice,
 			ImageOutputPrice: r.ImageOutputPrice,
 			PerRequestPrice:  r.PerRequestPrice,
 			Intervals:        intervals,
@@ -498,6 +502,7 @@ func (h *ChannelHandler) GetModelDefaultPricing(c *gin.Context) {
 		"output_price":       pricing.OutputPricePerToken,
 		"cache_write_price":  pricing.CacheCreationPricePerToken,
 		"cache_read_price":   pricing.CacheReadPricePerToken,
+		"image_input_price":  pricing.ImageInputPricePerToken,
 		"image_output_price": pricing.ImageOutputPricePerToken,
 	})
 }

backend/internal/handler/admin/channel_monitor_duplicate_test.go

@@ -0,0 +1,170 @@
+//go:build unit
+
+package admin
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+
+	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"
+)
+
+type duplicateChannelMonitorHandlerRepoStub struct {
+	service.ChannelMonitorRepository
+	source      *service.ChannelMonitor
+	byOperation map[string]*service.ChannelMonitor
+	createCalls int
+}
+
+func (r *duplicateChannelMonitorHandlerRepoStub) GetByID(_ context.Context, id int64) (*service.ChannelMonitor, error) {
+	if r.source == nil || r.source.ID != id {
+		return nil, service.ErrChannelMonitorNotFound
+	}
+	return r.source, nil
+}
+
+func (r *duplicateChannelMonitorHandlerRepoStub) Create(_ context.Context, monitor *service.ChannelMonitor) error {
+	r.createCalls++
+	monitor.ID = int64(100 + r.createCalls)
+	stored := *monitor
+	if stored.DuplicateOperationID != "" {
+		if r.byOperation == nil {
+			r.byOperation = make(map[string]*service.ChannelMonitor)
+		}
+		r.byOperation[stored.DuplicateOperationID] = &stored
+	}
+	return nil
+}
+
+func (r *duplicateChannelMonitorHandlerRepoStub) FindByDuplicateOperationID(_ context.Context, operationID string) (*service.ChannelMonitor, error) {
+	monitor := r.byOperation[operationID]
+	if monitor == nil {
+		return nil, nil
+	}
+	cloned := *monitor
+	return &cloned, nil
+}
+
+type duplicateChannelMonitorHandlerEncryptor struct{}
+
+func (duplicateChannelMonitorHandlerEncryptor) Encrypt(plaintext string) (string, error) {
+	return "ENC:" + plaintext, nil
+}
+
+func (duplicateChannelMonitorHandlerEncryptor) Decrypt(ciphertext string) (string, error) {
+	return strings.TrimPrefix(ciphertext, "ENC:"), nil
+}
+
+func setupDuplicateChannelMonitorRouter(t *testing.T) (*gin.Engine, *duplicateChannelMonitorHandlerRepoStub) {
+	t.Helper()
+	previousCoordinator := service.DefaultIdempotencyCoordinator()
+	service.SetDefaultIdempotencyCoordinator(service.NewIdempotencyCoordinator(
+		newMemoryIdempotencyRepoStub(),
+		service.DefaultIdempotencyConfig(),
+	))
+	t.Cleanup(func() { service.SetDefaultIdempotencyCoordinator(previousCoordinator) })
+
+	repo := &duplicateChannelMonitorHandlerRepoStub{
+		source: &service.ChannelMonitor{
+			ID:               42,
+			Name:             "primary",
+			Provider:         service.MonitorProviderOpenAI,
+			APIMode:          service.MonitorAPIModeResponses,
+			Endpoint:         "https://api.example.com",
+			APIKey:           "ENC:top-secret",
+			PrimaryModel:     "gpt-5.4-mini",
+			ExtraModels:      []string{"gpt-5.4"},
+			Enabled:          true,
+			IntervalSeconds:  60,
+			BodyOverrideMode: service.MonitorBodyOverrideModeOff,
+		},
+	}
+	monitorService := service.NewChannelMonitorService(repo, duplicateChannelMonitorHandlerEncryptor{})
+	handler := NewChannelMonitorHandler(monitorService)
+
+	gin.SetMode(gin.TestMode)
+	router := gin.New()
+	router.Use(func(c *gin.Context) {
+		c.Set(string(middleware2.ContextKeyUser), middleware2.AuthSubject{UserID: 77})
+		c.Next()
+	})
+	router.POST("/api/v1/admin/channel-monitors/:id/duplicate", handler.Duplicate)
+	return router, repo
+}
+
+func TestDuplicateChannelMonitorHandlerRedactsKeyAndReplaysRetry(t *testing.T) {
+	router, repo := setupDuplicateChannelMonitorRouter(t)
+
+	request := httptest.NewRequest(http.MethodPost, "/api/v1/admin/channel-monitors/42/duplicate", nil)
+	request.Header.Set("Idempotency-Key", "duplicate-channel-monitor-42")
+	first := httptest.NewRecorder()
+	router.ServeHTTP(first, request)
+
+	require.Equal(t, http.StatusOK, first.Code)
+	require.Equal(t, 1, repo.createCalls)
+	require.Contains(t, first.Body.String(), `"name":"primary (Copy)"`)
+	require.Contains(t, first.Body.String(), `"api_key_masked":"top-***"`)
+	require.Contains(t, first.Body.String(), `"created_by":77`)
+	require.Contains(t, first.Body.String(), `"enabled":false`)
+	require.NotContains(t, first.Body.String(), "top-secret")
+
+	retryRequest := httptest.NewRequest(http.MethodPost, "/api/v1/admin/channel-monitors/42/duplicate", nil)
+	retryRequest.Header.Set("Idempotency-Key", "duplicate-channel-monitor-42")
+	retry := httptest.NewRecorder()
+	router.ServeHTTP(retry, retryRequest)
+
+	require.Equal(t, http.StatusOK, retry.Code)
+	require.Equal(t, "true", retry.Header().Get("X-Idempotency-Replayed"))
+	require.Equal(t, 1, repo.createCalls)
+	require.JSONEq(t, first.Body.String(), retry.Body.String())
+}
+
+func TestDuplicateChannelMonitorHandlerRejectsInvalidID(t *testing.T) {
+	router, repo := setupDuplicateChannelMonitorRouter(t)
+	request := httptest.NewRequest(http.MethodPost, "/api/v1/admin/channel-monitors/not-a-number/duplicate", nil)
+	request.Header.Set("Idempotency-Key", "duplicate-channel-monitor-invalid")
+	recorder := httptest.NewRecorder()
+
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusBadRequest, recorder.Code)
+	require.Zero(t, repo.createCalls)
+}
+
+func TestDuplicateChannelMonitorHandlerRecoversAfterMarkSucceededFailure(t *testing.T) {
+	router, repo := setupDuplicateChannelMonitorRouter(t)
+	idempotencyRepo := &failOnceMarkSucceededRepo{
+		memoryIdempotencyRepoStub: newMemoryIdempotencyRepoStub(),
+		failNext:                  true,
+	}
+	service.SetDefaultIdempotencyCoordinator(service.NewIdempotencyCoordinator(
+		idempotencyRepo,
+		service.DefaultIdempotencyConfig(),
+	))
+
+	call := func() *httptest.ResponseRecorder {
+		request := httptest.NewRequest(http.MethodPost, "/api/v1/admin/channel-monitors/42/duplicate", nil)
+		request.Header.Set("Idempotency-Key", "duplicate-channel-monitor-recovery")
+		recorder := httptest.NewRecorder()
+		router.ServeHTTP(recorder, request)
+		return recorder
+	}
+
+	first := call()
+	second := call()
+
+	require.Equal(t, http.StatusOK, first.Code)
+	require.Equal(t, http.StatusOK, second.Code)
+	require.Equal(t, "true", first.Header().Get("X-Idempotency-Recovered"))
+	require.Equal(t, "true", second.Header().Get("X-Idempotency-Recovered"))
+	require.Equal(t, 1, repo.createCalls, "ambiguous retries must not repeat the create side effect")
+	require.Contains(t, second.Body.String(), `"id":101`)
+	require.Contains(t, second.Body.String(), `"api_key_masked":"top-***"`)
+	require.NotContains(t, second.Body.String(), "top-secret")
+}

backend/internal/handler/admin/channel_monitor_handler.go

@@ -1,6 +1,8 @@
 package admin
 
 import (
+	"context"
+	"log/slog"
 	"strconv"
 	"strings"
 	"time"
@@ -37,11 +39,11 @@ func NewChannelMonitorHandler(monitorService *service.ChannelMonitorService) *Ch
 
 type channelMonitorCreateRequest struct {
 	Name             string            `json:"name" binding:"required,max=100"`
-	Provider         string            `json:"provider" binding:"required,oneof=openai anthropic gemini"`
+	Provider         string            `json:"provider" binding:"required,oneof=openai anthropic gemini grok"`
 	APIMode          string            `json:"api_mode" binding:"omitempty,oneof=chat_completions responses"`
 	Endpoint         string            `json:"endpoint" binding:"required,max=500"`
 	APIKey           string            `json:"api_key" binding:"required,max=2000"`
-	PrimaryModel     string            `json:"primary_model" binding:"required,max=200"`
+	PrimaryModel     string            `json:"primary_model" binding:"max=200"`
 	ExtraModels      []string          `json:"extra_models"`
 	GroupName        string            `json:"group_name" binding:"max=100"`
 	Enabled          *bool             `json:"enabled"`
@@ -55,7 +57,7 @@ type channelMonitorCreateRequest struct {
 
 type channelMonitorUpdateRequest struct {
 	Name             *string            `json:"name" binding:"omitempty,max=100"`
-	Provider         *string            `json:"provider" binding:"omitempty,oneof=openai anthropic gemini"`
+	Provider         *string            `json:"provider" binding:"omitempty,oneof=openai anthropic gemini grok"`
 	APIMode          *string            `json:"api_mode" binding:"omitempty,oneof=chat_completions responses"`
 	Endpoint         *string            `json:"endpoint" binding:"omitempty,max=500"`
 	APIKey           *string            `json:"api_key" binding:"omitempty,max=2000"`
@@ -333,6 +335,63 @@ func (h *ChannelMonitorHandler) Create(c *gin.Context) {
 	response.Created(c, channelMonitorToResponse(m))
 }
 
+// Duplicate POST /api/v1/admin/channel-monitors/:id/duplicate
+func (h *ChannelMonitorHandler) Duplicate(c *gin.Context) {
+	id, ok := ParseChannelMonitorID(c)
+	if !ok {
+		return
+	}
+	subject, _ := middleware2.GetAuthSubjectFromContext(c)
+	actorScope := adminActorScope(c)
+
+	result, err := executeAdminIdempotent(
+		c,
+		"admin.channel_monitors.duplicate",
+		struct {
+			MonitorID int64 `json:"monitor_id"`
+		}{MonitorID: id},
+		service.DefaultWriteIdempotencyTTL(),
+		func(ctx context.Context) (any, error) {
+			monitor, err := h.monitorService.Duplicate(
+				ctx,
+				id,
+				subject.UserID,
+				actorScope,
+				c.GetHeader("Idempotency-Key"),
+			)
+			if err != nil {
+				return nil, err
+			}
+			return channelMonitorToResponse(monitor), nil
+		},
+	)
+	if err != nil {
+		reason := infraerrors.Reason(err)
+		if reason == infraerrors.Reason(service.ErrIdempotencyInProgress) || reason == infraerrors.Reason(service.ErrIdempotencyStoreUnavail) {
+			recovered, recoverErr := h.monitorService.RecoverDuplicate(
+				c.Request.Context(),
+				id,
+				actorScope,
+				c.GetHeader("Idempotency-Key"),
+			)
+			if recoverErr != nil {
+				slog.Warn("channel_monitor_duplicate_recovery_failed", "monitor_id", id, "actor_scope", actorScope, "reason", reason, "error", recoverErr)
+			} else if recovered != nil {
+				c.Header("X-Idempotency-Recovered", "true")
+				response.Success(c, channelMonitorToResponse(recovered))
+				return
+			}
+		}
+		response.ErrorFrom(c, err)
+		return
+	}
+
+	if result != nil && result.Replayed {
+		c.Header("X-Idempotency-Replayed", "true")
+	}
+	response.Success(c, result.Data)
+}
+
 // Update PUT /api/v1/admin/channel-monitors/:id
 func (h *ChannelMonitorHandler) Update(c *gin.Context) {
 	id, ok := ParseChannelMonitorID(c)

backend/internal/handler/admin/channel_monitor_template_handler.go

@@ -26,7 +26,7 @@ func NewChannelMonitorRequestTemplateHandler(templateService *service.ChannelMon
 
 type channelMonitorTemplateCreateRequest struct {
 	Name             string            `json:"name" binding:"required,max=100"`
-	Provider         string            `json:"provider" binding:"required,oneof=openai anthropic gemini"`
+	Provider         string            `json:"provider" binding:"required,oneof=openai anthropic gemini grok"`
 	APIMode          string            `json:"api_mode" binding:"omitempty,oneof=chat_completions responses"`
 	Description      string            `json:"description" binding:"max=500"`
 	ExtraHeaders     map[string]string `json:"extra_headers"`

backend/internal/handler/admin/grok_import_probe.go

@@ -0,0 +1,205 @@
+package admin
+
+import (
+	"context"
+	"log/slog"
+	"sync"
+	"time"
+
+	infraerrors "github.com/Wei-Shaw/sub2api/internal/pkg/errors"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+)
+
+const (
+	grokImportProbeConcurrency = 3
+	grokImportProbeTimeout     = 25 * time.Second
+)
+
+type grokUsageProber interface {
+	ProbeUsage(ctx context.Context, accountID int64) (*service.GrokQuotaProbeResult, error)
+}
+
+type grokImportProbeTask struct {
+	prober    grokUsageProber
+	accountID int64
+}
+
+type grokImportProbeScheduler struct {
+	mu          sync.Mutex
+	queue       []grokImportProbeTask
+	concurrency int
+	workers     int
+	maxWorkers  int
+	timeout     time.Duration
+}
+
+var defaultGrokImportProbeScheduler = newGrokImportProbeScheduler(
+	grokImportProbeConcurrency,
+	grokImportProbeTimeout,
+)
+
+func newGrokImportProbeScheduler(concurrency int, timeout time.Duration) *grokImportProbeScheduler {
+	if concurrency <= 0 {
+		concurrency = 1
+	}
+	if timeout <= 0 {
+		timeout = grokImportProbeTimeout
+	}
+	return &grokImportProbeScheduler{
+		concurrency: concurrency,
+		timeout:     timeout,
+	}
+}
+
+func (s *grokImportProbeScheduler) schedule(prober grokUsageProber, account *service.Account) {
+	if s == nil || prober == nil || account == nil || account.ID <= 0 {
+		return
+	}
+	if account.Platform != service.PlatformGrok || account.Type != service.AccountTypeOAuth {
+		return
+	}
+
+	s.mu.Lock()
+	s.queue = append(s.queue, grokImportProbeTask{prober: prober, accountID: account.ID})
+	if s.workers < s.concurrency {
+		s.workers++
+		if s.workers > s.maxWorkers {
+			s.maxWorkers = s.workers
+		}
+		go s.worker()
+	}
+	s.mu.Unlock()
+}
+
+func (s *grokImportProbeScheduler) worker() {
+	for {
+		task, ok := s.nextTask()
+		if !ok {
+			return
+		}
+		s.run(task.prober, task.accountID)
+	}
+}
+
+func (s *grokImportProbeScheduler) nextTask() (grokImportProbeTask, bool) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if len(s.queue) == 0 {
+		s.workers--
+		return grokImportProbeTask{}, false
+	}
+	task := s.queue[0]
+	s.queue[0] = grokImportProbeTask{}
+	s.queue = s.queue[1:]
+	if len(s.queue) == 0 {
+		s.queue = nil
+	}
+	return task, true
+}
+
+func (s *grokImportProbeScheduler) run(prober grokUsageProber, accountID int64) {
+	defer func() {
+		if recovered := recover(); recovered != nil {
+			slog.Error(
+				"grok_import_active_probe_panic",
+				"account_id", accountID,
+				"recovery_type", panicType(recovered),
+			)
+		}
+	}()
+
+	// Queue time is intentionally excluded: every imported account is probed,
+	// while this timeout only bounds the actual upstream probe execution.
+	ctx, cancel := context.WithTimeout(context.Background(), s.timeout)
+	defer cancel()
+	result, err := prober.ProbeUsage(ctx, accountID)
+	if err != nil {
+		slog.Warn(
+			"grok_import_active_probe_failed",
+			"account_id", accountID,
+			"status", infraerrors.Code(err),
+			"reason", infraerrors.Reason(err),
+		)
+		return
+	}
+	if result == nil {
+		slog.Warn(
+			"grok_import_active_probe_failed",
+			"account_id", accountID,
+			"reason", "empty_result",
+		)
+		return
+	}
+
+	slog.Info(
+		"grok_import_active_probe_completed",
+		"account_id", accountID,
+		"model", result.Model,
+		"status", result.StatusCode,
+		"headers_observed", result.HeadersObserved,
+	)
+}
+
+func panicType(value any) string {
+	switch value.(type) {
+	case string:
+		return "string"
+	case error:
+		return "error"
+	default:
+		return "unknown"
+	}
+}
+
+func (h *AccountHandler) scheduleGrokImportProbe(account *service.Account) {
+	if h == nil {
+		return
+	}
+	defaultGrokImportProbeScheduler.schedule(h.grokImportProber, account)
+}
+
+func (h *GrokOAuthHandler) scheduleGrokImportProbe(account *service.Account) {
+	if h == nil {
+		return
+	}
+	defaultGrokImportProbeScheduler.schedule(h.importProber, account)
+}
+
+// ProvideAccountHandler injects the Grok active prober for production while
+// keeping NewAccountHandler convenient for focused unit tests.
+func ProvideAccountHandler(
+	adminService service.AdminService,
+	oauthService *service.OAuthService,
+	openaiOAuthService *service.OpenAIOAuthService,
+	geminiOAuthService *service.GeminiOAuthService,
+	antigravityOAuthService *service.AntigravityOAuthService,
+	grokOAuthService service.GrokOAuthTokenService,
+	rateLimitService *service.RateLimitService,
+	accountUsageService *service.AccountUsageService,
+	accountTestService *service.AccountTestService,
+	concurrencyService *service.ConcurrencyService,
+	crsSyncService *service.CRSSyncService,
+	sessionLimitCache service.SessionLimitCache,
+	rpmCache service.RPMCache,
+	tokenCacheInvalidator service.TokenCacheInvalidator,
+	grokQuotaService *service.GrokQuotaService,
+) *AccountHandler {
+	handler := NewAccountHandler(
+		adminService,
+		oauthService,
+		openaiOAuthService,
+		geminiOAuthService,
+		antigravityOAuthService,
+		grokOAuthService,
+		rateLimitService,
+		accountUsageService,
+		accountTestService,
+		concurrencyService,
+		crsSyncService,
+		sessionLimitCache,
+		rpmCache,
+		tokenCacheInvalidator,
+	)
+	handler.grokImportProber = grokQuotaService
+	return handler
+}

backend/internal/handler/admin/grok_import_probe_handler_test.go

@@ -0,0 +1,119 @@
+//go:build unit
+
+package admin
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"sync"
+	"testing"
+	"time"
+
+	infraerrors "github.com/Wei-Shaw/sub2api/internal/pkg/errors"
+	"github.com/Wei-Shaw/sub2api/internal/pkg/xai"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/gin-gonic/gin"
+	"github.com/stretchr/testify/require"
+)
+
+type grokImportAdminService struct {
+	*stubAdminService
+	mu     sync.Mutex
+	nextID int64
+}
+
+func newGrokImportAdminService() *grokImportAdminService {
+	return &grokImportAdminService{
+		stubAdminService: newStubAdminService(),
+		nextID:           500,
+	}
+}
+
+func (s *grokImportAdminService) CreateAccount(_ context.Context, input *service.CreateAccountInput) (*service.Account, error) {
+	s.mu.Lock()
+	s.nextID++
+	id := s.nextID
+	s.mu.Unlock()
+	return &service.Account{
+		ID:          id,
+		Name:        input.Name,
+		Platform:    input.Platform,
+		Type:        input.Type,
+		Credentials: input.Credentials,
+		Extra:       input.Extra,
+		ProxyID:     input.ProxyID,
+		Concurrency: input.Concurrency,
+		Status:      service.StatusActive,
+		Schedulable: true,
+		CreatedAt:   time.Now(),
+		UpdatedAt:   time.Now(),
+	}, nil
+}
+
+type grokImportOAuthClientStub struct{}
+
+func (grokImportOAuthClientStub) ExchangeCode(context.Context, string, string, string, string, string) (*xai.TokenResponse, error) {
+	return &xai.TokenResponse{AccessToken: "access-token", RefreshToken: "refresh-token", ExpiresIn: 3600}, nil
+}
+
+func (grokImportOAuthClientStub) RefreshToken(context.Context, string, string, string) (*xai.TokenResponse, error) {
+	return &xai.TokenResponse{AccessToken: "access-token", RefreshToken: "refresh-token", ExpiresIn: 3600}, nil
+}
+
+func (grokImportOAuthClientStub) ConvertSSOToBuild(context.Context, string, string) (*xai.TokenResponse, error) {
+	return &xai.TokenResponse{AccessToken: "access-token", RefreshToken: "refresh-token", ExpiresIn: 3600}, nil
+}
+
+func TestGrokSSOBatchImportKeepsCreatedAccountsWhenOneAutomaticProbeFails(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	adminService := newGrokImportAdminService()
+	oauthService := service.NewGrokOAuthService(nil, grokImportOAuthClientStub{})
+	defer oauthService.Stop()
+	prober := newGrokImportProbeStub(3)
+	prober.failures[502] = infraerrors.New(502, "GROK_TEST_PROBE_FAILED", "sensitive-upstream-body")
+	handler := NewGrokOAuthHandler(oauthService, adminService, nil, nil)
+	handler.importProber = prober
+
+	router := gin.New()
+	router.POST("/api/v1/admin/grok/sso-to-oauth", handler.CreateAccountsFromSSO)
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(
+		http.MethodPost,
+		"/api/v1/admin/grok/sso-to-oauth",
+		strings.NewReader(`{"sso_tokens":["sso-one","sso-two","sso-three"]}`),
+	)
+	request.Header.Set("Content-Type", "application/json")
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusOK, recorder.Code)
+	require.Contains(t, recorder.Body.String(), `"created"`)
+	require.NotContains(t, recorder.Body.String(), `GROK_TEST_PROBE_FAILED`)
+	for i := 0; i < 3; i++ {
+		awaitGrokProbeSignal(t, prober.done)
+	}
+	calls, _, _ := prober.snapshot()
+	require.Equal(t, map[int64]int{501: 1, 502: 1, 503: 1}, calls)
+}
+
+func TestAccountCreateWithoutAutomaticGrokProbeServiceStillSucceeds(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	handler := NewAccountHandler(
+		newGrokImportAdminService(),
+		nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil,
+	)
+
+	router := gin.New()
+	router.POST("/api/v1/admin/accounts", handler.Create)
+	recorder := httptest.NewRecorder()
+	request := httptest.NewRequest(
+		http.MethodPost,
+		"/api/v1/admin/accounts",
+		strings.NewReader(`{"name":"grok-rt","platform":"grok","type":"oauth","credentials":{"refresh_token":"secret"}}`),
+	)
+	request.Header.Set("Content-Type", "application/json")
+	router.ServeHTTP(recorder, request)
+
+	require.Equal(t, http.StatusOK, recorder.Code)
+}

backend/internal/handler/admin/grok_import_probe_test.go

@@ -0,0 +1,232 @@
+//go:build unit
+
+package admin
+
+import (
+	"bytes"
+	"context"
+	"log/slog"
+	"sync"
+	"testing"
+	"time"
+
+	infraerrors "github.com/Wei-Shaw/sub2api/internal/pkg/errors"
+	"github.com/Wei-Shaw/sub2api/internal/service"
+	"github.com/stretchr/testify/require"
+)
+
+type grokImportProbeStub struct {
+	mu           sync.Mutex
+	calls        map[int64]int
+	failures     map[int64]error
+	active       int
+	maxActive    int
+	deadlineSeen bool
+	block        <-chan struct{}
+	started      chan int64
+	done         chan int64
+}
+
+func newGrokImportProbeStub(buffer int) *grokImportProbeStub {
+	return &grokImportProbeStub{
+		calls:    make(map[int64]int),
+		failures: make(map[int64]error),
+		started:  make(chan int64, buffer),
+		done:     make(chan int64, buffer),
+	}
+}
+
+func (s *grokImportProbeStub) ProbeUsage(ctx context.Context, accountID int64) (*service.GrokQuotaProbeResult, error) {
+	_, deadlineSeen := ctx.Deadline()
+	s.mu.Lock()
+	s.calls[accountID]++
+	s.active++
+	if s.active > s.maxActive {
+		s.maxActive = s.active
+	}
+	s.deadlineSeen = s.deadlineSeen || deadlineSeen
+	s.mu.Unlock()
+
+	s.started <- accountID
+	var ctxErr error
+	if s.block != nil {
+		select {
+		case <-s.block:
+		case <-ctx.Done():
+			ctxErr = ctx.Err()
+		}
+	}
+
+	s.mu.Lock()
+	s.active--
+	failure := s.failures[accountID]
+	s.mu.Unlock()
+	s.done <- accountID
+	if ctxErr != nil {
+		return nil, ctxErr
+	}
+	if failure != nil {
+		return nil, failure
+	}
+	return &service.GrokQuotaProbeResult{
+		Source:         "active_probe",
+		Model:          "grok-4.5",
+		StatusCode:     200,
+		ResetSupported: false,
+	}, nil
+}
+
+func (s *grokImportProbeStub) snapshot() (map[int64]int, int, bool) {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	calls := make(map[int64]int, len(s.calls))
+	for id, count := range s.calls {
+		calls[id] = count
+	}
+	return calls, s.maxActive, s.deadlineSeen
+}
+
+type grokImportProbeSchedulerTestSnapshot struct {
+	queued     int
+	workers    int
+	maxWorkers int
+}
+
+func snapshotGrokImportProbeScheduler(s *grokImportProbeScheduler) grokImportProbeSchedulerTestSnapshot {
+	if s == nil {
+		return grokImportProbeSchedulerTestSnapshot{}
+	}
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	return grokImportProbeSchedulerTestSnapshot{
+		queued:     len(s.queue),
+		workers:    s.workers,
+		maxWorkers: s.maxWorkers,
+	}
+}
+
+func newGrokOAuthImportAccount(id int64) *service.Account {
+	return &service.Account{
+		ID:       id,
+		Platform: service.PlatformGrok,
+		Type:     service.AccountTypeOAuth,
+	}
+}
+
+func awaitGrokProbeSignal(t *testing.T, signals <-chan int64) int64 {
+	t.Helper()
+	select {
+	case id := <-signals:
+		return id
+	case <-time.After(time.Second):
+		t.Fatal("timed out waiting for Grok import probe")
+		return 0
+	}
+}
+
+func TestGrokImportProbeSchedulerProbesSingleAccountOnce(t *testing.T) {
+	scheduler := newGrokImportProbeScheduler(1, time.Second)
+	prober := newGrokImportProbeStub(1)
+
+	scheduler.schedule(prober, newGrokOAuthImportAccount(101))
+	require.Equal(t, int64(101), awaitGrokProbeSignal(t, prober.done))
+
+	calls, maxActive, deadlineSeen := prober.snapshot()
+	require.Equal(t, map[int64]int{101: 1}, calls)
+	require.Equal(t, 1, maxActive)
+	require.True(t, deadlineSeen)
+	require.Eventually(t, func() bool {
+		snapshot := snapshotGrokImportProbeScheduler(scheduler)
+		return snapshot.queued == 0 && snapshot.workers == 0
+	}, time.Second, 10*time.Millisecond)
+}
+
+func TestGrokImportProbeSchedulerQueuesBatchWithoutPerTaskGoroutines(t *testing.T) {
+	const taskCount = 100
+	release := make(chan struct{})
+	scheduler := newGrokImportProbeScheduler(3, time.Second)
+	prober := newGrokImportProbeStub(taskCount)
+	prober.block = release
+	prober.failures[150] = infraerrors.New(502, "GROK_TEST_PROBE_FAILED", "sensitive-upstream-body")
+
+	for id := int64(101); id < 101+taskCount; id++ {
+		scheduler.schedule(prober, newGrokOAuthImportAccount(id))
+	}
+	for i := 0; i < 3; i++ {
+		awaitGrokProbeSignal(t, prober.started)
+	}
+	snapshot := snapshotGrokImportProbeScheduler(scheduler)
+	require.Equal(t, 97, snapshot.queued)
+	require.Equal(t, 3, snapshot.workers)
+	require.Equal(t, 3, snapshot.maxWorkers)
+	select {
+	case id := <-prober.started:
+		t.Fatalf("probe %d started before a concurrency slot was released", id)
+	case <-time.After(75 * time.Millisecond):
+	}
+	close(release)
+	for i := 0; i < taskCount; i++ {
+		awaitGrokProbeSignal(t, prober.done)
+	}
+
+	calls, maxActive, _ := prober.snapshot()
+	require.Len(t, calls, taskCount)
+	for id := int64(101); id < 101+taskCount; id++ {
+		require.Equal(t, 1, calls[id])
+	}
+	require.Equal(t, 3, maxActive)
+	require.Eventually(t, func() bool {
+		snapshot = snapshotGrokImportProbeScheduler(scheduler)
+		return snapshot.queued == 0 && snapshot.workers == 0
+	}, time.Second, 10*time.Millisecond)
+	require.Equal(t, 3, snapshot.maxWorkers)
+}
+
+func TestGrokImportProbeSchedulerTimeoutCancelsProbe(t *testing.T) {
+	neverRelease := make(chan struct{})
+	scheduler := newGrokImportProbeScheduler(1, 20*time.Millisecond)
+	prober := newGrokImportProbeStub(1)
+	prober.block = neverRelease
+
+	scheduler.schedule(prober, newGrokOAuthImportAccount(201))
+	require.Equal(t, int64(201), awaitGrokProbeSignal(t, prober.done))
+
+	calls, _, _ := prober.snapshot()
+	require.Equal(t, 1, calls[201])
+}
+
+func TestGrokImportProbeSchedulerSkipsMissingServiceAndNonGrokAccounts(t *testing.T) {
+	scheduler := newGrokImportProbeScheduler(1, time.Second)
+	prober := newGrokImportProbeStub(1)
+
+	scheduler.schedule(nil, newGrokOAuthImportAccount(301))
+	scheduler.schedule(prober, &service.Account{ID: 302, Platform: service.PlatformOpenAI, Type: service.AccountTypeOAuth})
+	scheduler.schedule(prober, &service.Account{ID: 303, Platform: service.PlatformGrok, Type: service.AccountTypeAPIKey})
+
+	select {
+	case id := <-prober.started:
+		t.Fatalf("unexpected probe for account %d", id)
+	case <-time.After(50 * time.Millisecond):
+	}
+	calls, _, _ := prober.snapshot()
+	require.Empty(t, calls)
+}
+
+func TestGrokImportProbeFailureLogDoesNotIncludeErrorMessage(t *testing.T) {
+	var logs bytes.Buffer
+	previousLogger := slog.Default()
+	slog.SetDefault(slog.New(slog.NewTextHandler(&logs, nil)))
+	defer slog.SetDefault(previousLogger)
+
+	scheduler := newGrokImportProbeScheduler(1, time.Second)
+	prober := newGrokImportProbeStub(1)
+	prober.failures[401] = infraerrors.New(502, "GROK_TEST_PROBE_FAILED", "refresh-token-secret")
+	scheduler.schedule(prober, newGrokOAuthImportAccount(401))
+	awaitGrokProbeSignal(t, prober.done)
+
+	require.Eventually(t, func() bool {
+		return bytes.Contains(logs.Bytes(), []byte("grok_import_active_probe_failed"))
+	}, time.Second, 10*time.Millisecond)
+	require.Contains(t, logs.String(), "GROK_TEST_PROBE_FAILED")
+	require.NotContains(t, logs.String(), "refresh-token-secret")
+}

backend/internal/handler/admin/grok_oauth_handler.go

@@ -1,31 +1,44 @@
 package admin
 
 import (
+	"context"
+	"fmt"
+	"log/slog"
 	"strconv"
 	"strings"
+	"sync"
+	"time"
 
 	"github.com/Wei-Shaw/sub2api/internal/handler/dto"
+	infraerrors "github.com/Wei-Shaw/sub2api/internal/pkg/errors"
 	"github.com/Wei-Shaw/sub2api/internal/pkg/response"
 	"github.com/Wei-Shaw/sub2api/internal/pkg/xai"
 	"github.com/Wei-Shaw/sub2api/internal/service"
 	"github.com/gin-gonic/gin"
 )
 
+const grokSSOImportConcurrency = 3
+
 type GrokOAuthHandler struct {
 	grokOAuthService *service.GrokOAuthService
 	adminService     service.AdminService
 	quotaService     *service.GrokQuotaService
+	importProber     grokUsageProber
+	reconciler       service.GrokOAuthReconciler
 }
 
 func NewGrokOAuthHandler(
 	grokOAuthService *service.GrokOAuthService,
 	adminService service.AdminService,
 	quotaService *service.GrokQuotaService,
+	reconciler service.GrokOAuthReconciler,
 ) *GrokOAuthHandler {
 	return &GrokOAuthHandler{
 		grokOAuthService: grokOAuthService,
 		adminService:     adminService,
 		quotaService:     quotaService,
+		importProber:     quotaService,
+		reconciler:       reconciler,
 	}
 }
 
@@ -151,6 +164,50 @@ func (h *GrokOAuthHandler) RefreshAccountToken(c *gin.Context) {
 	response.Success(c, dto.AccountFromService(updatedAccount))
 }
 
+type GrokOAuthReconcileRequest struct {
+	DryRun               *bool `json:"dry_run"`
+	Apply                bool  `json:"apply"`
+	AfterID              int64 `json:"after_id"`
+	Limit                int   `json:"limit"`
+	RefreshWindowSeconds int64 `json:"refresh_window_seconds"`
+}
+
+func (h *GrokOAuthHandler) ReconcileOAuthAccounts(c *gin.Context) {
+	var req GrokOAuthReconcileRequest
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.BadRequest(c, "Invalid request")
+		return
+	}
+	dryRun := true
+	if req.DryRun != nil {
+		dryRun = *req.DryRun
+	}
+	if req.Apply == dryRun {
+		response.ErrorFrom(c, service.ErrGrokOAuthReconcileMode)
+		return
+	}
+	if req.RefreshWindowSeconds < 0 || req.RefreshWindowSeconds > int64((24*time.Hour)/time.Second) {
+		response.ErrorFrom(c, service.ErrGrokOAuthReconcileWindow)
+		return
+	}
+	if h.reconciler == nil {
+		response.InternalError(c, "Grok OAuth reconciliation service is unavailable")
+		return
+	}
+	result, err := h.reconciler.ReconcileGrokOAuth(c.Request.Context(), service.GrokOAuthReconcileInput{
+		DryRun:        dryRun,
+		Apply:         req.Apply,
+		AfterID:       req.AfterID,
+		Limit:         req.Limit,
+		RefreshWindow: time.Duration(req.RefreshWindowSeconds) * time.Second,
+	})
+	if err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
+	response.Success(c, result)
+}
+
 func (h *GrokOAuthHandler) CreateAccountFromOAuth(c *gin.Context) {
 	var req struct {
 		SessionID   string  `json:"session_id" binding:"required"`
@@ -202,9 +259,243 @@ func (h *GrokOAuthHandler) CreateAccountFromOAuth(c *gin.Context) {
 		response.ErrorFrom(c, err)
 		return
 	}
+	h.scheduleGrokImportProbe(account)
 	response.Success(c, dto.AccountFromService(account))
 }
 
+type GrokSSOToOAuthRequest struct {
+	SSOTokens          []string       `json:"sso_tokens"`
+	SSOToken           string         `json:"sso_token"`
+	Name               string         `json:"name"`
+	Notes              *string        `json:"notes"`
+	ProxyID            *int64         `json:"proxy_id"`
+	GroupIDs           []int64        `json:"group_ids"`
+	Credentials        map[string]any `json:"credentials"`
+	Extra              map[string]any `json:"extra"`
+	Concurrency        int            `json:"concurrency"`
+	LoadFactor         *int           `json:"load_factor"`
+	Priority           int            `json:"priority"`
+	RateMultiplier     *float64       `json:"rate_multiplier"`
+	ExpiresAt          *int64         `json:"expires_at"`
+	AutoPauseOnExpired *bool          `json:"auto_pause_on_expired"`
+}
+
+type GrokSSOToOAuthItemResult struct {
+	Index   int          `json:"index"`
+	Name    string       `json:"name,omitempty"`
+	Email   string       `json:"email,omitempty"`
+	Account *dto.Account `json:"account,omitempty"`
+	Error   string       `json:"error,omitempty"`
+}
+
+type GrokSSOToOAuthResponse struct {
+	Created []GrokSSOToOAuthItemResult `json:"created"`
+	Failed  []GrokSSOToOAuthItemResult `json:"failed"`
+}
+
+type grokSSOImportJob struct {
+	index int
+	token string
+}
+
+type grokSSOImportWorkerResult struct {
+	created bool
+	item    GrokSSOToOAuthItemResult
+}
+
+func (h *GrokOAuthHandler) CreateAccountsFromSSO(c *gin.Context) {
+	var req GrokSSOToOAuthRequest
+	if err := c.ShouldBindJSON(&req); err != nil {
+		response.BadRequest(c, "Invalid request: "+err.Error())
+		return
+	}
+	tokens := normalizeSSOImportTokens(req.SSOTokens, req.SSOToken)
+	if len(tokens) == 0 {
+		response.BadRequest(c, "sso_tokens is required")
+		return
+	}
+
+	ctx := c.Request.Context()
+	workerCount := grokSSOImportConcurrency
+	if len(tokens) < workerCount {
+		workerCount = len(tokens)
+	}
+	jobs := make(chan grokSSOImportJob)
+	items := make([]grokSSOImportWorkerResult, len(tokens))
+	var wg sync.WaitGroup
+	for i := 0; i < workerCount; i++ {
+		wg.Add(1)
+		go func() {
+			defer wg.Done()
+			for job := range jobs {
+				items[job.index] = h.safeCreateAccountFromSSOToken(ctx, req, job.token, job.index+1, len(tokens))
+			}
+		}()
+	}
+	for i, token := range tokens {
+		jobs <- grokSSOImportJob{index: i, token: token}
+	}
+	close(jobs)
+	wg.Wait()
+
+	result := GrokSSOToOAuthResponse{
+		Created: make([]GrokSSOToOAuthItemResult, 0, len(tokens)),
+		Failed:  make([]GrokSSOToOAuthItemResult, 0),
+	}
+	for _, item := range items {
+		if item.created {
+			result.Created = append(result.Created, item.item)
+		} else {
+			result.Failed = append(result.Failed, item.item)
+		}
+	}
+	response.Success(c, result)
+}
+
+func (h *GrokOAuthHandler) safeCreateAccountFromSSOToken(ctx context.Context, req GrokSSOToOAuthRequest, token string, index, total int) (result grokSSOImportWorkerResult) {
+	defer func() {
+		if recovered := recover(); recovered != nil {
+			slog.Error("grok_sso_import_worker_panic", "index", index, "recover", recovered)
+			result = grokSSOImportWorkerResult{
+				item: GrokSSOToOAuthItemResult{
+					Index: index,
+					Error: fmt.Sprintf("internal worker panic: %v", recovered),
+				},
+			}
+		}
+	}()
+	return h.createAccountFromSSOToken(ctx, req, token, index, total)
+}
+
+func (h *GrokOAuthHandler) createAccountFromSSOToken(ctx context.Context, req GrokSSOToOAuthRequest, token string, index, total int) grokSSOImportWorkerResult {
+	tokenInfo, err := h.grokOAuthService.ConvertFromSSO(ctx, token, req.ProxyID)
+	if err != nil {
+		return grokSSOImportWorkerResult{item: GrokSSOToOAuthItemResult{Index: index, Error: grokSSOImportErrorMessage(err)}}
+	}
+
+	credentials := h.grokOAuthService.BuildAccountCredentials(tokenInfo)
+	credentials = service.MergeCredentials(cloneGrokSSOMap(req.Credentials), credentials)
+	name := grokSSOImportAccountName(req.Name, tokenInfo, index, total)
+	expiresAt, autoPauseOnExpired := grokSSOImportExpiry(req.ExpiresAt, req.AutoPauseOnExpired, tokenInfo)
+	account, err := h.adminService.CreateAccount(ctx, &service.CreateAccountInput{
+		Name:               name,
+		Notes:              req.Notes,
+		Platform:           service.PlatformGrok,
+		Type:               service.AccountTypeOAuth,
+		Credentials:        credentials,
+		Extra:              cloneGrokSSOMap(req.Extra),
+		ProxyID:            req.ProxyID,
+		Concurrency:        req.Concurrency,
+		LoadFactor:         req.LoadFactor,
+		Priority:           req.Priority,
+		RateMultiplier:     req.RateMultiplier,
+		GroupIDs:           append([]int64(nil), req.GroupIDs...),
+		ExpiresAt:          expiresAt,
+		AutoPauseOnExpired: autoPauseOnExpired,
+	})
+	if err != nil {
+		return grokSSOImportWorkerResult{item: GrokSSOToOAuthItemResult{Index: index, Name: name, Email: tokenInfo.Email, Error: grokSSOImportErrorMessage(err)}}
+	}
+	h.scheduleGrokImportProbe(account)
+	return grokSSOImportWorkerResult{
+		created: true,
+		item: GrokSSOToOAuthItemResult{
+			Index:   index,
+			Name:    name,
+			Email:   tokenInfo.Email,
+			Account: dto.AccountFromService(account),
+		},
+	}
+}
+
+func grokSSOImportExpiry(requestExpiresAt *int64, requestAutoPause *bool, tokenInfo *service.GrokTokenInfo) (*int64, *bool) {
+	if tokenInfo == nil || strings.TrimSpace(tokenInfo.RefreshToken) != "" || tokenInfo.ExpiresAt <= 0 {
+		return requestExpiresAt, requestAutoPause
+	}
+
+	expiresAt := tokenInfo.ExpiresAt
+	if requestExpiresAt != nil && *requestExpiresAt > 0 && *requestExpiresAt < expiresAt {
+		expiresAt = *requestExpiresAt
+	}
+	autoPause := true
+	return &expiresAt, &autoPause
+}
+
+func cloneGrokSSOMap(source map[string]any) map[string]any {
+	if source == nil {
+		return nil
+	}
+	clone := make(map[string]any, len(source))
+	for key, value := range source {
+		clone[key] = cloneGrokSSOValue(value)
+	}
+	return clone
+}
+
+func cloneGrokSSOValue(value any) any {
+	switch v := value.(type) {
+	case map[string]any:
+		return cloneGrokSSOMap(v)
+	case []any:
+		clone := make([]any, len(v))
+		for i, item := range v {
+			clone[i] = cloneGrokSSOValue(item)
+		}
+		return clone
+	default:
+		return value
+	}
+}
+
+func normalizeSSOImportTokens(tokens []string, single string) []string {
+	items := make([]string, 0, len(tokens)+1)
+	if strings.TrimSpace(single) != "" {
+		items = append(items, single)
+	}
+	items = append(items, tokens...)
+	seen := make(map[string]struct{}, len(items))
+	result := make([]string, 0, len(items))
+	for _, item := range items {
+		parts := strings.Split(strings.NewReplacer(",", "\n", "\r", "\n").Replace(item), "\n")
+		for _, token := range parts {
+			if token = xai.NormalizeSSOToken(token); token == "" {
+				continue
+			}
+			if _, ok := seen[token]; ok {
+				continue
+			}
+			seen[token] = struct{}{}
+			result = append(result, token)
+		}
+	}
+	return result
+}
+
+func grokSSOImportAccountName(base string, tokenInfo *service.GrokTokenInfo, index, total int) string {
+	base = strings.TrimSpace(base)
+	if base == "" && tokenInfo != nil {
+		base = strings.TrimSpace(tokenInfo.Email)
+	}
+	if base == "" {
+		base = "Grok OAuth Account"
+	}
+	if total > 1 {
+		return base + " #" + strconv.Itoa(index)
+	}
+	return base
+}
+
+func grokSSOImportErrorMessage(err error) string {
+	status := infraerrors.FromError(err)
+	if status == nil {
+		return ""
+	}
+	if status.Reason != "" {
+		return status.Reason + ": " + status.Message
+	}
+	return status.Message
+}
+
 func (h *GrokOAuthHandler) QueryQuota(c *gin.Context) {
 	accountID, err := strconv.ParseInt(c.Param("id"), 10, 64)
 	if err != nil {
@@ -215,7 +506,7 @@ func (h *GrokOAuthHandler) QueryQuota(c *gin.Context) {
 		response.BadRequest(c, "grok quota service is not enabled")
 		return
 	}
-	result, err := h.quotaService.ProbeUsage(c.Request.Context(), accountID)
+	result, err := h.quotaService.QueryQuota(c.Request.Context(), accountID)
 	if err != nil {
 		response.ErrorFrom(c, err)
 		return

backend/internal/handler/admin/grok_oauth_handler_test.go

@@ -8,6 +8,7 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"strings"
+	"sync"
 	"testing"
 	"time"
 
@@ -41,17 +42,48 @@ func (r *grokQuotaHandlerAccountRepo) UpdateExtra(_ context.Context, id int64, u
 }
 
 type grokQuotaHandlerUpstream struct {
-	resp     *http.Response
-	lastReq  *http.Request
-	lastBody []byte
+	mu       sync.Mutex
+	requests []*http.Request
+	bodies   [][]byte
+}
+
+type grokOAuthReconcilerStub struct {
+	input  service.GrokOAuthReconcileInput
+	calls  int
+	result *service.GrokOAuthReconcileResult
+	err    error
+}
+
+func (s *grokOAuthReconcilerStub) ReconcileGrokOAuth(_ context.Context, input service.GrokOAuthReconcileInput) (*service.GrokOAuthReconcileResult, error) {
+	s.calls++
+	s.input = input
+	return s.result, s.err
 }
 
 func (u *grokQuotaHandlerUpstream) Do(req *http.Request, _ string, _ int64, _ int) (*http.Response, error) {
-	u.lastReq = req
+	var body []byte
 	if req.Body != nil {
-		u.lastBody, _ = io.ReadAll(req.Body)
+		body, _ = io.ReadAll(req.Body)
 	}
-	return u.resp, nil
+	u.mu.Lock()
+	u.requests = append(u.requests, req)
+	u.bodies = append(u.bodies, body)
+	u.mu.Unlock()
+	if req.URL.Path == "/v1/responses" {
+		return &http.Response{
+			StatusCode: http.StatusOK,
+			Header: http.Header{
+				"X-Ratelimit-Limit-Requests":     []string{"10"},
+				"X-Ratelimit-Remaining-Requests": []string{"8"},
+			},
+			Body: io.NopCloser(strings.NewReader(`{"id":"resp_probe"}`)),
+		}, nil
+	}
+	payload := `{"config":{"billingPeriodStart":"2026-07-01T00:00:00Z","billingPeriodEnd":"2026-08-01T00:00:00Z"}}`
+	if req.URL.RawQuery == "format=credits" {
+		payload = `{"config":{"currentPeriod":{"type":"WEEKLY","start":"2026-07-09T03:25:00Z","end":"2026-07-16T03:25:00Z"}}}`
+	}
+	return &http.Response{StatusCode: http.StatusOK, Header: make(http.Header), Body: io.NopCloser(strings.NewReader(payload))}, nil
 }
 
 func (u *grokQuotaHandlerUpstream) DoWithTLS(
@@ -71,22 +103,18 @@ func TestGrokOAuthHandlerQueryQuotaProbesUpstream(t *testing.T) {
 		ID:          42,
 		Platform:    service.PlatformGrok,
 		Type:        service.AccountTypeOAuth,
+		Status:      service.StatusActive,
+		Schedulable: true,
 		Concurrency: 1,
 		Credentials: map[string]any{
-			"access_token": "access-token",
-			"expires_at":   time.Now().Add(time.Hour).UTC().Format(time.RFC3339),
+			"access_token":  "access-token",
+			"refresh_token": "refresh-token",
+			"expires_at":    time.Now().Add(2 * time.Hour).UTC().Format(time.RFC3339),
 		},
 	}}
-	upstream := &grokQuotaHandlerUpstream{resp: &http.Response{
-		StatusCode: http.StatusOK,
-		Header: http.Header{
-			"X-Ratelimit-Limit-Requests":     []string{"10"},
-			"X-Ratelimit-Remaining-Requests": []string{"8"},
-		},
-		Body: io.NopCloser(strings.NewReader(`{"id":"resp_probe"}`)),
-	}}
-	quotaService := service.NewGrokQuotaService(repo, nil, service.NewGrokTokenProvider(repo, nil), upstream)
-	handler := NewGrokOAuthHandler(nil, nil, quotaService)
+	upstream := &grokQuotaHandlerUpstream{}
+	quotaService := service.NewGrokQuotaService(repo, nil, service.NewGrokTokenProvider(repo, nil), upstream, nil)
+	handler := NewGrokOAuthHandler(nil, nil, quotaService, nil)
 
 	router := gin.New()
 	router.GET("/api/v1/admin/grok/accounts/:id/quota", handler.QueryQuota)
@@ -95,12 +123,23 @@ func TestGrokOAuthHandlerQueryQuotaProbesUpstream(t *testing.T) {
 	router.ServeHTTP(rec, req)
 
 	require.Equal(t, http.StatusOK, rec.Code)
-	require.Contains(t, rec.Body.String(), `"source":"active_probe"`)
+	require.Contains(t, rec.Body.String(), `"source":"hybrid_probe"`)
+	require.Contains(t, rec.Body.String(), `"billing":`)
+	require.Contains(t, rec.Body.String(), `"snapshot":`)
 	require.Contains(t, rec.Body.String(), `"headers_observed":true`)
 	require.NotContains(t, rec.Body.String(), "access-token")
-	require.Equal(t, xai.DefaultBaseURL+"/responses", upstream.lastReq.URL.String())
-	require.Equal(t, "Bearer access-token", upstream.lastReq.Header.Get("Authorization"))
-	require.Contains(t, string(upstream.lastBody), `"store":false`)
+	upstream.mu.Lock()
+	requests := append([]*http.Request(nil), upstream.requests...)
+	bodies := append([][]byte(nil), upstream.bodies...)
+	upstream.mu.Unlock()
+	require.Len(t, requests, 3)
+	for i, upstreamReq := range requests {
+		require.Equal(t, "Bearer access-token", upstreamReq.Header.Get("Authorization"))
+		if upstreamReq.URL.String() == xai.DefaultCLIBaseURL+"/responses" {
+			require.Contains(t, string(bodies[i]), `"model":"grok-4.5"`)
+			require.Contains(t, string(bodies[i]), `"store":false`)
+		}
+	}
 	require.NotNil(t, repo.updates[42])
 }
 
@@ -112,8 +151,8 @@ func TestGrokOAuthHandlerResetQuotaReturnsUnsupported(t *testing.T) {
 		Platform: service.PlatformGrok,
 		Type:     service.AccountTypeOAuth,
 	}}
-	quotaService := service.NewGrokQuotaService(repo, nil, nil, nil)
-	handler := NewGrokOAuthHandler(nil, nil, quotaService)
+	quotaService := service.NewGrokQuotaService(repo, nil, nil, nil, nil)
+	handler := NewGrokOAuthHandler(nil, nil, quotaService, nil)
 
 	router := gin.New()
 	router.POST("/api/v1/admin/grok/accounts/:id/reset-quota", handler.ResetQuota)
@@ -131,7 +170,7 @@ func TestGrokOAuthHandlerRuntimeSanityDoesNotExposeSecrets(t *testing.T) {
 	t.Setenv(xai.EnvBaseURL, "http://127.0.0.1:8080/v1?access_token=secret")
 	t.Setenv(xai.EnvClientID, "client-secret-like-value")
 
-	handler := NewGrokOAuthHandler(nil, nil, nil)
+	handler := NewGrokOAuthHandler(nil, nil, nil, nil)
 	router := gin.New()
 	router.GET("/api/v1/admin/grok/runtime-sanity", handler.RuntimeSanity)
 	rec := httptest.NewRecorder()
@@ -145,3 +184,117 @@ func TestGrokOAuthHandlerRuntimeSanityDoesNotExposeSecrets(t *testing.T) {
 	require.NotContains(t, rec.Body.String(), "secret")
 	require.NotContains(t, rec.Body.String(), "client-secret-like-value")
 }
+
+func TestGrokSSOImportExpiryUsesTokenExpiryWithoutRefreshToken(t *testing.T) {
+	tokenExpiry := time.Now().Add(6 * time.Hour).Unix()
+	expiresAt, autoPause := grokSSOImportExpiry(nil, nil, &service.GrokTokenInfo{
+		ExpiresAt: tokenExpiry,
+	})
+
+	require.NotNil(t, expiresAt)
+	require.Equal(t, tokenExpiry, *expiresAt)
+	require.NotNil(t, autoPause)
+	require.True(t, *autoPause)
+}
+
+func TestGrokSSOImportExpiryUsesEarlierRequestedExpiryWithoutRefreshToken(t *testing.T) {
+	requestedExpiry := time.Now().Add(2 * time.Hour).Unix()
+	tokenExpiry := time.Now().Add(6 * time.Hour).Unix()
+	requestedAutoPause := false
+	expiresAt, autoPause := grokSSOImportExpiry(&requestedExpiry, &requestedAutoPause, &service.GrokTokenInfo{
+		ExpiresAt: tokenExpiry,
+	})
+
+	require.NotNil(t, expiresAt)
+	require.Equal(t, requestedExpiry, *expiresAt)
+	require.NotNil(t, autoPause)
+	require.True(t, *autoPause)
+}
+
+func TestGrokSSOImportExpiryPreservesRequestSettingsWithRefreshToken(t *testing.T) {
+	requestedExpiry := time.Now().Add(2 * time.Hour).Unix()
+	requestedAutoPause := false
+	expiresAt, autoPause := grokSSOImportExpiry(&requestedExpiry, &requestedAutoPause, &service.GrokTokenInfo{
+		RefreshToken: "refresh-token",
+		ExpiresAt:    time.Now().Add(6 * time.Hour).Unix(),
+	})
+
+	require.Same(t, &requestedExpiry, expiresAt)
+	require.Same(t, &requestedAutoPause, autoPause)
+}
+
+func TestGrokSSOImportWorkerRecoversPanic(t *testing.T) {
+	h := &GrokOAuthHandler{}
+	result := h.safeCreateAccountFromSSOToken(context.Background(), GrokSSOToOAuthRequest{}, "token", 2, 3)
+	// Without a service, createAccountFromSSOToken would panic on nil service access.
+	// Recovery must convert that into a failed item and keep the worker alive.
+	require.False(t, result.created)
+	require.Equal(t, 2, result.item.Index)
+	require.Contains(t, result.item.Error, "internal worker panic")
+}
+
+func TestGrokOAuthHandlerReconcileDefaultsToDryRun(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	reconciler := &grokOAuthReconcilerStub{result: &service.GrokOAuthReconcileResult{
+		DryRun:      true,
+		Scanned:     2,
+		Actionable:  1,
+		WouldBlock:  1,
+		Items:       []service.GrokOAuthReconcileItem{{AccountID: 42, Reason: service.GrokOAuthReconcileReasonMissingRefreshToken, Action: service.GrokOAuthReconcileActionBlock, Outcome: service.GrokOAuthReconcileOutcomePlanned}},
+		NextAfterID: 0,
+	}}
+	handler := NewGrokOAuthHandler(nil, nil, nil, reconciler)
+	router := gin.New()
+	router.POST("/api/v1/admin/grok/oauth/reconcile", handler.ReconcileOAuthAccounts)
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequest(http.MethodPost, "/api/v1/admin/grok/oauth/reconcile", strings.NewReader(`{}`))
+	req.Header.Set("Content-Type", "application/json")
+
+	router.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Equal(t, 1, reconciler.calls)
+	require.True(t, reconciler.input.DryRun)
+	require.False(t, reconciler.input.Apply)
+	require.Contains(t, rec.Body.String(), `"reason":"missing_refresh_token"`)
+	require.NotContains(t, rec.Body.String(), `"refresh_token":`)
+	require.NotContains(t, rec.Body.String(), `"access_token":`)
+}
+
+func TestGrokOAuthHandlerReconcileRequiresExplicitApply(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	reconciler := &grokOAuthReconcilerStub{}
+	handler := NewGrokOAuthHandler(nil, nil, nil, reconciler)
+	router := gin.New()
+	router.POST("/api/v1/admin/grok/oauth/reconcile", handler.ReconcileOAuthAccounts)
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequest(http.MethodPost, "/api/v1/admin/grok/oauth/reconcile", strings.NewReader(`{"dry_run":false}`))
+	req.Header.Set("Content-Type", "application/json")
+
+	router.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusBadRequest, rec.Code)
+	require.Zero(t, reconciler.calls)
+	require.NotContains(t, rec.Body.String(), "credentials")
+}
+
+func TestGrokOAuthHandlerReconcileExplicitApply(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	reconciler := &grokOAuthReconcilerStub{result: &service.GrokOAuthReconcileResult{DryRun: false, Refreshed: 1}}
+	handler := NewGrokOAuthHandler(nil, nil, nil, reconciler)
+	router := gin.New()
+	router.POST("/api/v1/admin/grok/oauth/reconcile", handler.ReconcileOAuthAccounts)
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequest(http.MethodPost, "/api/v1/admin/grok/oauth/reconcile", strings.NewReader(`{"apply":true,"dry_run":false,"after_id":10,"limit":25,"refresh_window_seconds":3600}`))
+	req.Header.Set("Content-Type", "application/json")
+
+	router.ServeHTTP(rec, req)
+
+	require.Equal(t, http.StatusOK, rec.Code)
+	require.Equal(t, 1, reconciler.calls)
+	require.True(t, reconciler.input.Apply)
+	require.False(t, reconciler.input.DryRun)
+	require.Equal(t, int64(10), reconciler.input.AfterID)
+	require.Equal(t, 25, reconciler.input.Limit)
+	require.Equal(t, time.Hour, reconciler.input.RefreshWindow)
+}

backend/internal/handler/admin/group_handler.go

@@ -110,6 +110,7 @@ type CreateGroupRequest struct {
 	VideoPrice480P                  *float64 `json:"video_price_480p"`
 	VideoPrice720P                  *float64 `json:"video_price_720p"`
 	VideoPrice1080P                 *float64 `json:"video_price_1080p"`
+	WebSearchPricePerCall           *float64 `json:"web_search_price_per_call"`
 	ClaudeCodeOnly                  bool     `json:"claude_code_only"`
 	FallbackGroupID                 *int64   `json:"fallback_group_id"`
 	FallbackGroupIDOnInvalidRequest *int64   `json:"fallback_group_id_on_invalid_request"`
@@ -163,6 +164,7 @@ type UpdateGroupRequest struct {
 	VideoPrice480P                  *float64 `json:"video_price_480p"`
 	VideoPrice720P                  *float64 `json:"video_price_720p"`
 	VideoPrice1080P                 *float64 `json:"video_price_1080p"`
+	WebSearchPricePerCall           *float64 `json:"web_search_price_per_call"`
 	ClaudeCodeOnly                  *bool    `json:"claude_code_only"`
 	FallbackGroupID                 *int64   `json:"fallback_group_id"`
 	FallbackGroupIDOnInvalidRequest *int64   `json:"fallback_group_id_on_invalid_request"`
@@ -334,6 +336,7 @@ func (h *GroupHandler) Create(c *gin.Context) {
 		VideoPrice480P:                  req.VideoPrice480P,
 		VideoPrice720P:                  req.VideoPrice720P,
 		VideoPrice1080P:                 req.VideoPrice1080P,
+		WebSearchPricePerCall:           req.WebSearchPricePerCall,
 		ClaudeCodeOnly:                  req.ClaudeCodeOnly,
 		FallbackGroupID:                 req.FallbackGroupID,
 		FallbackGroupIDOnInvalidRequest: req.FallbackGroupIDOnInvalidRequest,
@@ -402,6 +405,7 @@ func (h *GroupHandler) Update(c *gin.Context) {
 		VideoPrice480P:                  req.VideoPrice480P,
 		VideoPrice720P:                  req.VideoPrice720P,
 		VideoPrice1080P:                 req.VideoPrice1080P,
+		WebSearchPricePerCall:           req.WebSearchPricePerCall,
 		ClaudeCodeOnly:                  req.ClaudeCodeOnly,
 		FallbackGroupID:                 req.FallbackGroupID,
 		FallbackGroupIDOnInvalidRequest: req.FallbackGroupIDOnInvalidRequest,

backend/internal/handler/admin/idempotency_helper.go

@@ -37,14 +37,9 @@ func executeAdminIdempotent(
 		return &service.IdempotencyExecuteResult{Data: data}, nil
 	}
 
-	actorScope := "admin:0"
-	if subject, ok := middleware2.GetAuthSubjectFromContext(c); ok {
-		actorScope = "admin:" + strconv.FormatInt(subject.UserID, 10)
-	}
-
 	return coordinator.Execute(c.Request.Context(), service.IdempotencyExecuteOptions{
 		Scope:          scope,
-		ActorScope:     actorScope,
+		ActorScope:     adminActorScope(c),
 		Method:         c.Request.Method,
 		Route:          c.FullPath(),
 		IdempotencyKey: c.GetHeader("Idempotency-Key"),
@@ -54,6 +49,14 @@ func executeAdminIdempotent(
 	}, execute)
 }
 
+func adminActorScope(c *gin.Context) string {
+	actorScope := "admin:0"
+	if subject, ok := middleware2.GetAuthSubjectFromContext(c); ok {
+		actorScope = "admin:" + strconv.FormatInt(subject.UserID, 10)
+	}
+	return actorScope
+}
+
 func executeAdminIdempotentJSON(
 	c *gin.Context,
 	scope string,

backend/internal/handler/admin/openai_oauth_handler.go

@@ -304,6 +304,10 @@ func (h *OpenAIOAuthHandler) CreateAccountFromCodexPAT(c *gin.Context) {
 		response.BadRequest(c, "Invalid request: "+err.Error())
 		return
 	}
+	if err := service.ValidateOpenAILongContextBillingExtra(service.PlatformOpenAI, req.Extra); err != nil {
+		response.ErrorFrom(c, err)
+		return
+	}
 	if req.Concurrency != nil && *req.Concurrency < 0 {
 		response.BadRequest(c, "concurrency must be >= 0")
 		return

backend/internal/handler/admin/ops_handler.go

@@ -386,8 +386,8 @@ func (h *OpsHandler) ListRequestErrorUpstreamErrors(c *gin.Context) {
 		filter.EndTime = &endTime
 	}
 	filter.View = "all"
-	filter.Phase = "upstream"
-	// 上游错误列表需含 status<400 的 recovered 行,显式豁免客户端可见守卫。
+	filter.ErrorPhasesAny = []string{"upstream", "account_auth"}
+	// Provider-health list includes recovered inference and credential rows.
 	filter.IncludeRecoveredUpstream = true
 	filter.Owner = "provider"
 	filter.Source = strings.TrimSpace(c.Query("error_source"))
@@ -470,8 +470,8 @@ func (h *OpsHandler) ListUpstreamErrors(c *gin.Context) {
 	}
 
 	filter.View = parseOpsViewParam(c)
-	filter.Phase = "upstream"
-	// 上游错误列表需含 status<400 的 recovered 行,显式豁免客户端可见守卫。
+	filter.ErrorPhasesAny = []string{"upstream", "account_auth"}
+	// Provider-health list includes recovered inference and credential rows.
 	filter.IncludeRecoveredUpstream = true
 	filter.Owner = "provider"
 	filter.Source = strings.TrimSpace(c.Query("error_source"))

backend/internal/handler/admin/ops_system_log_handler.go

@@ -15,6 +15,7 @@ import (
 type opsSystemLogCleanupRequest struct {
 	StartTime string `json:"start_time"`
 	EndTime   string `json:"end_time"`
+	Host      string `json:"host"`
 
 	Level           string `json:"level"`
 	Component       string `json:"component"`
@@ -56,6 +57,7 @@ func (h *OpsHandler) ListSystemLogs(c *gin.Context) {
 		PageSize:        pageSize,
 		StartTime:       &start,
 		EndTime:         &end,
+		Host:            strings.TrimSpace(c.Query("host")),
 		Level:           strings.TrimSpace(c.Query("level")),
 		Component:       strings.TrimSpace(c.Query("component")),
 		RequestID:       strings.TrimSpace(c.Query("request_id")),
@@ -153,6 +155,7 @@ func (h *OpsHandler) CleanupSystemLogs(c *gin.Context) {
 	filter := &service.OpsSystemLogCleanupFilter{
 		StartTime:       start,
 		EndTime:         end,
+		Host:            strings.TrimSpace(req.Host),
 		Level:           strings.TrimSpace(req.Level),
 		Component:       strings.TrimSpace(req.Component),
 		RequestID:       strings.TrimSpace(req.RequestID),

backend/internal/handler/admin/ops_system_log_handler_test.go

@@ -2,6 +2,7 @@ package admin
 
 import (
 	"bytes"
+	"context"
 	"encoding/json"
 	"net/http"
 	"net/http/httptest"
@@ -19,6 +20,26 @@ type responseEnvelope struct {
 	Data    json.RawMessage `json:"data"`
 }
 
+type opsSystemLogCaptureRepo struct {
+	service.OpsRepository
+	listFilter    *service.OpsSystemLogFilter
+	cleanupFilter *service.OpsSystemLogCleanupFilter
+}
+
+func (r *opsSystemLogCaptureRepo) ListSystemLogs(_ context.Context, filter *service.OpsSystemLogFilter) (*service.OpsSystemLogList, error) {
+	r.listFilter = filter
+	return &service.OpsSystemLogList{Logs: []*service.OpsSystemLog{}, Page: filter.Page, PageSize: filter.PageSize}, nil
+}
+
+func (r *opsSystemLogCaptureRepo) DeleteSystemLogs(_ context.Context, filter *service.OpsSystemLogCleanupFilter) (int64, error) {
+	r.cleanupFilter = filter
+	return 1, nil
+}
+
+func (r *opsSystemLogCaptureRepo) InsertSystemLogCleanupAudit(_ context.Context, _ *service.OpsSystemLogCleanupAudit) error {
+	return nil
+}
+
 func newOpsSystemLogTestRouter(handler *OpsHandler, withUser bool) *gin.Engine {
 	gin.SetMode(gin.TestMode)
 	r := gin.New()
@@ -121,6 +142,23 @@ func TestOpsSystemLogHandler_ListSuccess(t *testing.T) {
 	}
 }
 
+func TestOpsSystemLogHandler_ListAcceptsHost(t *testing.T) {
+	repo := &opsSystemLogCaptureRepo{}
+	svc := service.NewOpsService(repo, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	h := NewOpsHandler(svc)
+	r := newOpsSystemLogTestRouter(h, false)
+
+	w := httptest.NewRecorder()
+	req := httptest.NewRequest(http.MethodGet, "/logs?host=api-node-1", nil)
+	r.ServeHTTP(w, req)
+	if w.Code != http.StatusOK {
+		t.Fatalf("status=%d, want 200", w.Code)
+	}
+	if repo.listFilter == nil || repo.listFilter.Host != "api-node-1" {
+		t.Fatalf("host filter = %+v, want api-node-1", repo.listFilter)
+	}
+}
+
 func TestOpsSystemLogHandler_CleanupUnauthorized(t *testing.T) {
 	svc := service.NewOpsService(nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	h := NewOpsHandler(svc)
@@ -205,6 +243,24 @@ func TestOpsSystemLogHandler_CleanupAcceptsAPIKeyID(t *testing.T) {
 	}
 }
 
+func TestOpsSystemLogHandler_CleanupAcceptsHost(t *testing.T) {
+	repo := &opsSystemLogCaptureRepo{}
+	svc := service.NewOpsService(repo, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
+	h := NewOpsHandler(svc)
+	r := newOpsSystemLogTestRouter(h, true)
+
+	w := httptest.NewRecorder()
+	req := httptest.NewRequest(http.MethodPost, "/logs/cleanup", bytes.NewBufferString(`{"host":"api-node-1"}`))
+	req.Header.Set("Content-Type", "application/json")
+	r.ServeHTTP(w, req)
+	if w.Code != http.StatusOK {
+		t.Fatalf("status=%d, want 200", w.Code)
+	}
+	if repo.cleanupFilter == nil || repo.cleanupFilter.Host != "api-node-1" {
+		t.Fatalf("host filter = %+v, want api-node-1", repo.cleanupFilter)
+	}
+}
+
 func TestOpsSystemLogHandler_CleanupInvalidAPIKeyID(t *testing.T) {
 	svc := service.NewOpsService(nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil)
 	h := NewOpsHandler(svc)

backend/internal/handler/admin/ops_ws_handler.go

@@ -16,6 +16,7 @@ import (
 	"time"
 
 	"github.com/Wei-Shaw/sub2api/internal/pkg/logger"
+	servermiddleware "github.com/Wei-Shaw/sub2api/internal/server/middleware"
 	"github.com/Wei-Shaw/sub2api/internal/service"
 	"github.com/gin-gonic/gin"
 	"github.com/gorilla/websocket"
@@ -323,7 +324,7 @@ func (h *OpsHandler) QPSWSHandler(c *gin.Context) {
 	// If realtime monitoring is disabled, prefer a successful WS upgrade followed by a clean close
 	// with a deterministic close code. This prevents clients from spinning on 404/1006 reconnect loops.
 	if !h.opsService.IsRealtimeMonitoringEnabled(c.Request.Context()) {
-		conn, err := upgrader.Upgrade(c.Writer, c.Request, nil)
+		conn, err := upgrader.Upgrade(c.Writer, c.Request, servermiddleware.ServerTimingResponseHeader(c))
 		if err != nil {
 			c.JSON(http.StatusNotFound, gin.H{"error": "ops realtime monitoring is disabled"})
 			return
@@ -358,7 +359,7 @@ func (h *OpsHandler) QPSWSHandler(c *gin.Context) {
 		defer releaseOpsWSIPSlot(clientIP)
 	}
 
-	conn, err := upgrader.Upgrade(c.Writer, c.Request, nil)
+	conn, err := upgrader.Upgrade(c.Writer, c.Request, servermiddleware.ServerTimingResponseHeader(c))
 	if err != nil {
 		logger.LegacyPrintf("handler.admin.ops_ws", "[OpsWS] upgrade failed: %v", err)
 		return

backend/internal/handler/admin/setting_handler.go

@@ -123,6 +123,8 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 		InvitationCodeEnabled:                                  settings.InvitationCodeEnabled,
 		TotpEnabled:                                            settings.TotpEnabled,
 		TotpEncryptionKeyConfigured:                            h.settingService.IsTotpEncryptionKeyConfigured(),
+		SessionBindingEnabled:                                  settings.SessionBindingEnabled,
+		AuditLogRetentionDays:                                  settings.AuditLogRetentionDays,
 		LoginAgreementEnabled:                                  settings.LoginAgreementEnabled,
 		LoginAgreementMode:                                     settings.LoginAgreementMode,
 		LoginAgreementUpdatedAt:                                settings.LoginAgreementUpdatedAt,
@@ -229,6 +231,7 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 		AffiliateRebateFreezeHours:                             settings.AffiliateRebateFreezeHours,
 		AffiliateRebateDurationDays:                            settings.AffiliateRebateDurationDays,
 		AffiliateRebatePerInviteeCap:                           settings.AffiliateRebatePerInviteeCap,
+		AdminRechargeRebateEnabled:                             settings.AdminRechargeRebateEnabled,
 		DefaultUserRPMLimit:                                    settings.DefaultUserRPMLimit,
 		DefaultSubscriptions:                                   defaultSubscriptions,
 		EnableModelFallback:                                    settings.EnableModelFallback,
@@ -268,6 +271,8 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 		PaymentVisibleMethodWxpaySource:                        settings.PaymentVisibleMethodWxpaySource,
 		PaymentVisibleMethodAlipayEnabled:                      settings.PaymentVisibleMethodAlipayEnabled,
 		PaymentVisibleMethodWxpayEnabled:                       settings.PaymentVisibleMethodWxpayEnabled,
+		OpenAILowUpstreamRatePriorityEnabled:                   settings.OpenAILowUpstreamRatePriorityEnabled,
+		OpenAIOAuthSchedulingRateMultiplier:                    settings.OpenAIOAuthSchedulingRateMultiplier,
 		OpenAIAdvancedSchedulerEnabled:                         settings.OpenAIAdvancedSchedulerEnabled,
 		OpenAIAdvancedSchedulerStickyWeightedEnabled:           settings.OpenAIAdvancedSchedulerStickyWeightedEnabled,
 		OpenAIAdvancedSchedulerSubscriptionPriorityEnabled:     settings.OpenAIAdvancedSchedulerSubscriptionPriorityEnabled,
@@ -279,6 +284,7 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 		OpenAIAdvancedSchedulerWeightTTFT:                      settings.OpenAIAdvancedSchedulerWeightTTFT,
 		OpenAIAdvancedSchedulerWeightReset:                     settings.OpenAIAdvancedSchedulerWeightReset,
 		OpenAIAdvancedSchedulerWeightQuotaHeadroom:             settings.OpenAIAdvancedSchedulerWeightQuotaHeadroom,
+		OpenAIAdvancedSchedulerWeightUpstreamCost:              settings.OpenAIAdvancedSchedulerWeightUpstreamCost,
 		OpenAIAdvancedSchedulerWeightPreviousResponse:          settings.OpenAIAdvancedSchedulerWeightPreviousResponse,
 		OpenAIAdvancedSchedulerWeightSessionSticky:             settings.OpenAIAdvancedSchedulerWeightSessionSticky,
 		OpenAIAdvancedSchedulerEffectiveLBTopK:                 settings.OpenAIAdvancedSchedulerEffectiveLBTopK,
@@ -289,6 +295,7 @@ func (h *SettingHandler) GetSettings(c *gin.Context) {
 		OpenAIAdvancedSchedulerEffectiveWeightTTFT:             settings.OpenAIAdvancedSchedulerEffectiveWeightTTFT,
 		OpenAIAdvancedSchedulerEffectiveWeightReset:            settings.OpenAIAdvancedSchedulerEffectiveWeightReset,
 		OpenAIAdvancedSchedulerEffectiveWeightQuotaHeadroom:    settings.OpenAIAdvancedSchedulerEffectiveWeightQuotaHeadroom,
+		OpenAIAdvancedSchedulerEffectiveWeightUpstreamCost:     settings.OpenAIAdvancedSchedulerEffectiveWeightUpstreamCost,
 		OpenAIAdvancedSchedulerEffectiveWeightPreviousResponse: settings.OpenAIAdvancedSchedulerEffectiveWeightPreviousResponse,
 		OpenAIAdvancedSchedulerEffectiveWeightSessionSticky:    settings.OpenAIAdvancedSchedulerEffectiveWeightSessionSticky,
 		BalanceLowNotifyEnabled:                                settings.BalanceLowNotifyEnabled,

backend/internal/handler/admin/setting_handler_audit.go

@@ -308,6 +308,9 @@ func diffSettings(before *service.SystemSettings, after *service.SystemSettings,
 	if before.AffiliateRebatePerInviteeCap != after.AffiliateRebatePerInviteeCap {
 		changed = append(changed, "affiliate_rebate_per_invitee_cap")
 	}
+	if before.AdminRechargeRebateEnabled != after.AdminRechargeRebateEnabled {
+		changed = append(changed, "affiliate_admin_recharge_enabled")
+	}
 	if !equalDefaultSubscriptions(before.DefaultSubscriptions, after.DefaultSubscriptions) {
 		changed = append(changed, "default_subscriptions")
 	}
@@ -437,6 +440,12 @@ func diffSettings(before *service.SystemSettings, after *service.SystemSettings,
 	if before.PaymentVisibleMethodWxpayEnabled != after.PaymentVisibleMethodWxpayEnabled {
 		changed = append(changed, "payment_visible_method_wxpay_enabled")
 	}
+	if before.OpenAILowUpstreamRatePriorityEnabled != after.OpenAILowUpstreamRatePriorityEnabled {
+		changed = append(changed, "openai_low_upstream_rate_priority_enabled")
+	}
+	if before.OpenAIOAuthSchedulingRateMultiplier != after.OpenAIOAuthSchedulingRateMultiplier {
+		changed = append(changed, "openai_oauth_scheduling_rate_multiplier")
+	}
 	if before.OpenAIAdvancedSchedulerEnabled != after.OpenAIAdvancedSchedulerEnabled {
 		changed = append(changed, "openai_advanced_scheduler_enabled")
 	}
@@ -470,6 +479,9 @@ func diffSettings(before *service.SystemSettings, after *service.SystemSettings,
 	if before.OpenAIAdvancedSchedulerWeightQuotaHeadroom != after.OpenAIAdvancedSchedulerWeightQuotaHeadroom {
 		changed = append(changed, "openai_advanced_scheduler_weight_quota_headroom")
 	}
+	if before.OpenAIAdvancedSchedulerWeightUpstreamCost != after.OpenAIAdvancedSchedulerWeightUpstreamCost {
+		changed = append(changed, "openai_advanced_scheduler_weight_upstream_cost")
+	}
 	if before.OpenAIAdvancedSchedulerWeightPreviousResponse != after.OpenAIAdvancedSchedulerWeightPreviousResponse {
 		changed = append(changed, "openai_advanced_scheduler_weight_previous_response")
 	}

backend/internal/handler/admin/setting_handler_auth_source_defaults_test.go

@@ -223,6 +223,7 @@ func TestSettingHandler_UpdateSettings_PersistsPaymentVisibleMethodsAndAdvancedS
 		"payment_visible_method_alipay_enabled":                   true,
 		"payment_visible_method_wxpay_enabled":                    false,
 		"openai_advanced_scheduler_enabled":                       true,
+		"openai_oauth_scheduling_rate_multiplier":                 0.05,
 		"openai_advanced_scheduler_subscription_priority_enabled": true,
 	}
 	rawBody, err := json.Marshal(body)
@@ -241,6 +242,7 @@ func TestSettingHandler_UpdateSettings_PersistsPaymentVisibleMethodsAndAdvancedS
 	require.Equal(t, "true", repo.values[service.SettingPaymentVisibleMethodAlipayEnabled])
 	require.Equal(t, "false", repo.values[service.SettingPaymentVisibleMethodWxpayEnabled])
 	require.Equal(t, "true", repo.values["openai_advanced_scheduler_enabled"])
+	require.Equal(t, "0.05", repo.values[service.SettingKeyOpenAIOAuthSchedulingRateMultiplier])
 	require.Equal(t, "true", repo.values[service.SettingKeyOpenAIAdvancedSchedulerSubscriptionPriorityEnabled])
 
 	var resp response.Response
@@ -252,6 +254,7 @@ func TestSettingHandler_UpdateSettings_PersistsPaymentVisibleMethodsAndAdvancedS
 	require.Equal(t, true, data["payment_visible_method_alipay_enabled"])
 	require.Equal(t, false, data["payment_visible_method_wxpay_enabled"])
 	require.Equal(t, true, data["openai_advanced_scheduler_enabled"])
+	require.Equal(t, 0.05, data["openai_oauth_scheduling_rate_multiplier"])
 	require.Equal(t, true, data["openai_advanced_scheduler_subscription_priority_enabled"])
 }
 

backend/internal/handler/admin/setting_handler_update.go

@@ -26,7 +26,9 @@ type UpdateSettingsRequest struct {
 	PasswordResetEnabled             bool                         `json:"password_reset_enabled"`
 	FrontendURL                      string                       `json:"frontend_url"`
 	InvitationCodeEnabled            bool                         `json:"invitation_code_enabled"`
-	TotpEnabled                      bool                         `json:"totp_enabled"` // TOTP 双因素认证
+	TotpEnabled                      bool                         `json:"totp_enabled"`             // TOTP 双因素认证
+	SessionBindingEnabled            bool                         `json:"session_binding_enabled"`  // 会话 IP/UA 绑定
+	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"`
@@ -149,6 +151,7 @@ type UpdateSettingsRequest struct {
 	AffiliateRebateFreezeHours                *int                              `json:"affiliate_rebate_freeze_hours"`
 	AffiliateRebateDurationDays               *int                              `json:"affiliate_rebate_duration_days"`
 	AffiliateRebatePerInviteeCap              *float64                          `json:"affiliate_rebate_per_invitee_cap"`
+	AdminRechargeRebateEnabled                *bool                             `json:"affiliate_admin_recharge_enabled"`
 	DefaultUserRPMLimit                       int                               `json:"default_user_rpm_limit"`
 	DefaultSubscriptions                      []dto.DefaultSubscriptionSetting  `json:"default_subscriptions"`
 	AuthSourceDefaultEmailBalance             *float64                          `json:"auth_source_default_email_balance"`
@@ -242,19 +245,22 @@ type UpdateSettingsRequest struct {
 	PaymentVisibleMethodWxpayEnabled  *bool   `json:"payment_visible_method_wxpay_enabled"`
 
 	// OpenAI account scheduling
-	OpenAIAdvancedSchedulerEnabled                     *bool   `json:"openai_advanced_scheduler_enabled"`
-	OpenAIAdvancedSchedulerStickyWeightedEnabled       *bool   `json:"openai_advanced_scheduler_sticky_weighted_enabled"`
-	OpenAIAdvancedSchedulerSubscriptionPriorityEnabled *bool   `json:"openai_advanced_scheduler_subscription_priority_enabled"`
-	OpenAIAdvancedSchedulerLBTopK                      *string `json:"openai_advanced_scheduler_lb_top_k"`
-	OpenAIAdvancedSchedulerWeightPriority              *string `json:"openai_advanced_scheduler_weight_priority"`
-	OpenAIAdvancedSchedulerWeightLoad                  *string `json:"openai_advanced_scheduler_weight_load"`
-	OpenAIAdvancedSchedulerWeightQueue                 *string `json:"openai_advanced_scheduler_weight_queue"`
-	OpenAIAdvancedSchedulerWeightErrorRate             *string `json:"openai_advanced_scheduler_weight_error_rate"`
-	OpenAIAdvancedSchedulerWeightTTFT                  *string `json:"openai_advanced_scheduler_weight_ttft"`
-	OpenAIAdvancedSchedulerWeightReset                 *string `json:"openai_advanced_scheduler_weight_reset"`
-	OpenAIAdvancedSchedulerWeightQuotaHeadroom         *string `json:"openai_advanced_scheduler_weight_quota_headroom"`
-	OpenAIAdvancedSchedulerWeightPreviousResponse      *string `json:"openai_advanced_scheduler_weight_previous_response"`
-	OpenAIAdvancedSchedulerWeightSessionSticky         *string `json:"openai_advanced_scheduler_weight_session_sticky"`
+	OpenAILowUpstreamRatePriorityEnabled               *bool    `json:"openai_low_upstream_rate_priority_enabled"`
+	OpenAIOAuthSchedulingRateMultiplier                *float64 `json:"openai_oauth_scheduling_rate_multiplier"`
+	OpenAIAdvancedSchedulerEnabled                     *bool    `json:"openai_advanced_scheduler_enabled"`
+	OpenAIAdvancedSchedulerStickyWeightedEnabled       *bool    `json:"openai_advanced_scheduler_sticky_weighted_enabled"`
+	OpenAIAdvancedSchedulerSubscriptionPriorityEnabled *bool    `json:"openai_advanced_scheduler_subscription_priority_enabled"`
+	OpenAIAdvancedSchedulerLBTopK                      *string  `json:"openai_advanced_scheduler_lb_top_k"`
+	OpenAIAdvancedSchedulerWeightPriority              *string  `json:"openai_advanced_scheduler_weight_priority"`
+	OpenAIAdvancedSchedulerWeightLoad                  *string  `json:"openai_advanced_scheduler_weight_load"`
+	OpenAIAdvancedSchedulerWeightQueue                 *string  `json:"openai_advanced_scheduler_weight_queue"`
+	OpenAIAdvancedSchedulerWeightErrorRate             *string  `json:"openai_advanced_scheduler_weight_error_rate"`
+	OpenAIAdvancedSchedulerWeightTTFT                  *string  `json:"openai_advanced_scheduler_weight_ttft"`
+	OpenAIAdvancedSchedulerWeightReset                 *string  `json:"openai_advanced_scheduler_weight_reset"`
+	OpenAIAdvancedSchedulerWeightQuotaHeadroom         *string  `json:"openai_advanced_scheduler_weight_quota_headroom"`
+	OpenAIAdvancedSchedulerWeightUpstreamCost          *string  `json:"openai_advanced_scheduler_weight_upstream_cost"`
+	OpenAIAdvancedSchedulerWeightPreviousResponse      *string  `json:"openai_advanced_scheduler_weight_previous_response"`
+	OpenAIAdvancedSchedulerWeightSessionSticky         *string  `json:"openai_advanced_scheduler_weight_session_sticky"`
 
 	// 余额不足提醒
 	BalanceLowNotifyEnabled         *bool                   `json:"balance_low_notify_enabled"`
@@ -391,6 +397,10 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 	if affiliateRebatePerInviteeCap < 0 {
 		affiliateRebatePerInviteeCap = service.AffiliateRebatePerInviteeCapDefault
 	}
+	adminRechargeRebateEnabled := previousSettings.AdminRechargeRebateEnabled
+	if req.AdminRechargeRebateEnabled != nil {
+		adminRechargeRebateEnabled = *req.AdminRechargeRebateEnabled
+	}
 	// 通用表格配置：兼容旧客户端未传字段时保留当前值。
 	if req.TableDefaultPageSize <= 0 {
 		req.TableDefaultPageSize = previousSettings.TableDefaultPageSize
@@ -1171,6 +1181,8 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		FrontendURL:                      req.FrontendURL,
 		InvitationCodeEnabled:            req.InvitationCodeEnabled,
 		TotpEnabled:                      req.TotpEnabled,
+		SessionBindingEnabled:            req.SessionBindingEnabled,
+		AuditLogRetentionDays:            req.AuditLogRetentionDays,
 		LoginAgreementEnabled:            req.LoginAgreementEnabled,
 		LoginAgreementMode:               loginAgreementMode,
 		LoginAgreementUpdatedAt:          loginAgreementUpdatedAt,
@@ -1279,6 +1291,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		AffiliateRebateFreezeHours:             affiliateRebateFreezeHours,
 		AffiliateRebateDurationDays:            affiliateRebateDurationDays,
 		AffiliateRebatePerInviteeCap:           affiliateRebatePerInviteeCap,
+		AdminRechargeRebateEnabled:             adminRechargeRebateEnabled,
 		DefaultUserRPMLimit:                    req.DefaultUserRPMLimit,
 		DefaultSubscriptions:                   defaultSubscriptions,
 		EnableModelFallback:                    req.EnableModelFallback,
@@ -1423,6 +1436,18 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 			}
 			return previousSettings.PaymentVisibleMethodWxpayEnabled
 		}(),
+		OpenAILowUpstreamRatePriorityEnabled: func() bool {
+			if req.OpenAILowUpstreamRatePriorityEnabled != nil {
+				return *req.OpenAILowUpstreamRatePriorityEnabled
+			}
+			return previousSettings.OpenAILowUpstreamRatePriorityEnabled
+		}(),
+		OpenAIOAuthSchedulingRateMultiplier: func() float64 {
+			if req.OpenAIOAuthSchedulingRateMultiplier != nil {
+				return *req.OpenAIOAuthSchedulingRateMultiplier
+			}
+			return previousSettings.OpenAIOAuthSchedulingRateMultiplier
+		}(),
 		OpenAIAdvancedSchedulerEnabled: func() bool {
 			if req.OpenAIAdvancedSchedulerEnabled != nil {
 				return *req.OpenAIAdvancedSchedulerEnabled
@@ -1449,6 +1474,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		OpenAIAdvancedSchedulerWeightTTFT:             stringSetting(req.OpenAIAdvancedSchedulerWeightTTFT, previousSettings.OpenAIAdvancedSchedulerWeightTTFT),
 		OpenAIAdvancedSchedulerWeightReset:            stringSetting(req.OpenAIAdvancedSchedulerWeightReset, previousSettings.OpenAIAdvancedSchedulerWeightReset),
 		OpenAIAdvancedSchedulerWeightQuotaHeadroom:    stringSetting(req.OpenAIAdvancedSchedulerWeightQuotaHeadroom, previousSettings.OpenAIAdvancedSchedulerWeightQuotaHeadroom),
+		OpenAIAdvancedSchedulerWeightUpstreamCost:     stringSetting(req.OpenAIAdvancedSchedulerWeightUpstreamCost, previousSettings.OpenAIAdvancedSchedulerWeightUpstreamCost),
 		OpenAIAdvancedSchedulerWeightPreviousResponse: stringSetting(req.OpenAIAdvancedSchedulerWeightPreviousResponse, previousSettings.OpenAIAdvancedSchedulerWeightPreviousResponse),
 		OpenAIAdvancedSchedulerWeightSessionSticky:    stringSetting(req.OpenAIAdvancedSchedulerWeightSessionSticky, previousSettings.OpenAIAdvancedSchedulerWeightSessionSticky),
 		BalanceLowNotifyEnabled: func() bool {
@@ -1683,6 +1709,8 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		InvitationCodeEnabled:                                  updatedSettings.InvitationCodeEnabled,
 		TotpEnabled:                                            updatedSettings.TotpEnabled,
 		TotpEncryptionKeyConfigured:                            h.settingService.IsTotpEncryptionKeyConfigured(),
+		SessionBindingEnabled:                                  updatedSettings.SessionBindingEnabled,
+		AuditLogRetentionDays:                                  updatedSettings.AuditLogRetentionDays,
 		LoginAgreementEnabled:                                  updatedSettings.LoginAgreementEnabled,
 		LoginAgreementMode:                                     updatedSettings.LoginAgreementMode,
 		LoginAgreementUpdatedAt:                                updatedSettings.LoginAgreementUpdatedAt,
@@ -1786,6 +1814,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		AffiliateRebateFreezeHours:                             updatedSettings.AffiliateRebateFreezeHours,
 		AffiliateRebateDurationDays:                            updatedSettings.AffiliateRebateDurationDays,
 		AffiliateRebatePerInviteeCap:                           updatedSettings.AffiliateRebatePerInviteeCap,
+		AdminRechargeRebateEnabled:                             updatedSettings.AdminRechargeRebateEnabled,
 		DefaultUserRPMLimit:                                    updatedSettings.DefaultUserRPMLimit,
 		DefaultSubscriptions:                                   updatedDefaultSubscriptions,
 		EnableModelFallback:                                    updatedSettings.EnableModelFallback,
@@ -1824,6 +1853,8 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		PaymentVisibleMethodWxpaySource:                        updatedSettings.PaymentVisibleMethodWxpaySource,
 		PaymentVisibleMethodAlipayEnabled:                      updatedSettings.PaymentVisibleMethodAlipayEnabled,
 		PaymentVisibleMethodWxpayEnabled:                       updatedSettings.PaymentVisibleMethodWxpayEnabled,
+		OpenAILowUpstreamRatePriorityEnabled:                   updatedSettings.OpenAILowUpstreamRatePriorityEnabled,
+		OpenAIOAuthSchedulingRateMultiplier:                    updatedSettings.OpenAIOAuthSchedulingRateMultiplier,
 		OpenAIAdvancedSchedulerEnabled:                         updatedSettings.OpenAIAdvancedSchedulerEnabled,
 		OpenAIAdvancedSchedulerStickyWeightedEnabled:           updatedSettings.OpenAIAdvancedSchedulerStickyWeightedEnabled,
 		OpenAIAdvancedSchedulerSubscriptionPriorityEnabled:     updatedSettings.OpenAIAdvancedSchedulerSubscriptionPriorityEnabled,
@@ -1835,6 +1866,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		OpenAIAdvancedSchedulerWeightTTFT:                      updatedSettings.OpenAIAdvancedSchedulerWeightTTFT,
 		OpenAIAdvancedSchedulerWeightReset:                     updatedSettings.OpenAIAdvancedSchedulerWeightReset,
 		OpenAIAdvancedSchedulerWeightQuotaHeadroom:             updatedSettings.OpenAIAdvancedSchedulerWeightQuotaHeadroom,
+		OpenAIAdvancedSchedulerWeightUpstreamCost:              updatedSettings.OpenAIAdvancedSchedulerWeightUpstreamCost,
 		OpenAIAdvancedSchedulerWeightPreviousResponse:          updatedSettings.OpenAIAdvancedSchedulerWeightPreviousResponse,
 		OpenAIAdvancedSchedulerWeightSessionSticky:             updatedSettings.OpenAIAdvancedSchedulerWeightSessionSticky,
 		OpenAIAdvancedSchedulerEffectiveLBTopK:                 updatedSettings.OpenAIAdvancedSchedulerEffectiveLBTopK,
@@ -1845,6 +1877,7 @@ func (h *SettingHandler) UpdateSettings(c *gin.Context) {
 		OpenAIAdvancedSchedulerEffectiveWeightTTFT:             updatedSettings.OpenAIAdvancedSchedulerEffectiveWeightTTFT,
 		OpenAIAdvancedSchedulerEffectiveWeightReset:            updatedSettings.OpenAIAdvancedSchedulerEffectiveWeightReset,
 		OpenAIAdvancedSchedulerEffectiveWeightQuotaHeadroom:    updatedSettings.OpenAIAdvancedSchedulerEffectiveWeightQuotaHeadroom,
+		OpenAIAdvancedSchedulerEffectiveWeightUpstreamCost:     updatedSettings.OpenAIAdvancedSchedulerEffectiveWeightUpstreamCost,
 		OpenAIAdvancedSchedulerEffectiveWeightPreviousResponse: updatedSettings.OpenAIAdvancedSchedulerEffectiveWeightPreviousResponse,
 		OpenAIAdvancedSchedulerEffectiveWeightSessionSticky:    updatedSettings.OpenAIAdvancedSchedulerEffectiveWeightSessionSticky,
 		BalanceLowNotifyEnabled:                                updatedSettings.BalanceLowNotifyEnabled,

backend/internal/handler/admin/user_handler.go

@@ -13,6 +13,7 @@ import (
 	"github.com/Wei-Shaw/sub2api/internal/handler/dto"
 	"github.com/Wei-Shaw/sub2api/internal/handler/quotaview"
 	"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"
@@ -30,6 +31,8 @@ type UserHandler struct {
 	concurrencyService    *service.ConcurrencyService
 	userPlatformQuotaRepo service.UserPlatformQuotaRepository // T13 admin quota view
 	billingCache          service.BillingCache                // T17/T18 缓存失效（PUT/POST 路径）
+	totpService           *service.TotpService                // 角色提升为管理员的 step-up 门控
+	userService           *service.UserService
 }
 
 // NewUserHandler creates a new admin user handler
@@ -38,12 +41,16 @@ func NewUserHandler(
 	concurrencyService *service.ConcurrencyService,
 	userPlatformQuotaRepo service.UserPlatformQuotaRepository,
 	billingCache service.BillingCache,
+	totpService *service.TotpService,
+	userService *service.UserService,
 ) *UserHandler {
 	return &UserHandler{
 		adminService:          adminService,
 		concurrencyService:    concurrencyService,
 		userPlatformQuotaRepo: userPlatformQuotaRepo,
 		billingCache:          billingCache,
+		totpService:           totpService,
+		userService:           userService,
 	}
 }
 
@@ -266,6 +273,13 @@ func (h *UserHandler) Create(c *gin.Context) {
 		return
 	}
 
+	// 创建管理员账号属权限敏感操作：需最近完成 step-up 2FA 验证。
+	if req.Role == service.RoleAdmin {
+		if !middleware.EnforceStepUp(c, h.totpService, h.userService) {
+			return
+		}
+	}
+
 	user, err := h.adminService.CreateUser(c.Request.Context(), &service.CreateUserInput{
 		Email:         req.Email,
 		Password:      req.Password,
@@ -308,6 +322,21 @@ func (h *UserHandler) Update(c *gin.Context) {
 		return
 	}
 
+	// 把普通用户提升为管理员属权限敏感操作：需最近完成 step-up 2FA 验证。
+	// 目标已是管理员时（前端编辑表单总是携带 role）不触发，避免日常编辑被打断。
+	if req.Role == service.RoleAdmin {
+		target, err := h.adminService.GetUser(c.Request.Context(), userID)
+		if err != nil {
+			response.ErrorFrom(c, err)
+			return
+		}
+		if target.Role != service.RoleAdmin {
+			if !middleware.EnforceStepUp(c, h.totpService, h.userService) {
+				return
+			}
+		}
+	}
+
 	// 使用指针类型直接传递，nil 表示未提供该字段
 	user, err := h.adminService.UpdateUser(c.Request.Context(), userID, &service.UpdateUserInput{
 		Email:         req.Email,

backend/internal/handler/admin/user_handler_activity_test.go

@@ -35,7 +35,7 @@ func TestUserHandlerListIncludesActivityFieldsAndSortParams(t *testing.T) {
 			UpdatedAt:    lastLoginAt,
 		},
 	}
-	handler := NewUserHandler(adminSvc, nil, nil, nil)
+	handler := NewUserHandler(adminSvc, nil, nil, nil, nil, nil)
 
 	recorder := httptest.NewRecorder()
 	c, _ := gin.CreateTestContext(recorder)
@@ -89,7 +89,7 @@ func TestUserHandlerGetByIDIncludesActivityFields(t *testing.T) {
 			UpdatedAt:    lastLoginAt,
 		},
 	}
-	handler := NewUserHandler(adminSvc, nil, nil, nil)
+	handler := NewUserHandler(adminSvc, nil, nil, nil, nil, nil)
 
 	recorder := httptest.NewRecorder()
 	c, _ := gin.CreateTestContext(recorder)

backend/internal/handler/admin/user_handler_get_deleted_test.go

@@ -26,7 +26,7 @@ func (s *getByIDAdminStub) GetUserIncludeDeleted(_ context.Context, id int64) (*
 func setupGetByIDRouter(svc service.AdminService) *gin.Engine {
 	gin.SetMode(gin.TestMode)
 	r := gin.New()
-	h := NewUserHandler(svc, nil, nil, nil)
+	h := NewUserHandler(svc, nil, nil, nil, nil, nil)
 	r.GET("/admin/users/:id", h.GetByID)
 	return r
 }

backend/internal/handler/admin/user_handler_list_apikey_group_test.go

@@ -39,7 +39,7 @@ func TestAdminUserList_ParsesAPIKeyGroupID(t *testing.T) {
 		t.Run(tc.name, func(t *testing.T) {
 			stub := &listUsersFilterStub{AdminService: newStubAdminService()}
 			r := gin.New()
-			h := NewUserHandler(stub, nil, nil, nil)
+			h := NewUserHandler(stub, nil, nil, nil, nil, nil)
 			r.GET("/admin/users", h.List)
 
 			w := httptest.NewRecorder()

backend/internal/handler/admin/user_handler_role_stepup_test.go

@@ -0,0 +1,86 @@
+package admin
+
+import (
+	"bytes"
+	"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"
+)
+
+// 角色提升为管理员的 step-up 门控条件测试。
+// 测试环境不注入认证上下文，因此门控一旦触发会以 401 中止；
+// 借此区分「触发了 step-up 校验」与「直接放行到业务层（200）」。
+func setupRoleStepUpRouter(t *testing.T) (*gin.Engine, *stubAdminService) {
+	t.Helper()
+	gin.SetMode(gin.TestMode)
+	router := gin.New()
+	adminSvc := newStubAdminService()
+	// 追加一个已是管理员的目标用户，验证「目标已是 admin 不触发门控」。
+	adminSvc.users = append(adminSvc.users, service.User{
+		ID:     2,
+		Email:  "admin@example.com",
+		Role:   service.RoleAdmin,
+		Status: service.StatusActive,
+	})
+
+	h := NewUserHandler(adminSvc, nil, nil, nil, nil, nil)
+	router.POST("/api/v1/admin/users", h.Create)
+	router.PUT("/api/v1/admin/users/:id", h.Update)
+	return router, adminSvc
+}
+
+func doJSON(t *testing.T, router *gin.Engine, method, path string, payload map[string]any) *httptest.ResponseRecorder {
+	t.Helper()
+	body, err := json.Marshal(payload)
+	require.NoError(t, err)
+	rec := httptest.NewRecorder()
+	req := httptest.NewRequest(method, path, bytes.NewReader(body))
+	req.Header.Set("Content-Type", "application/json")
+	router.ServeHTTP(rec, req)
+	return rec
+}
+
+func TestUpdateUserPromoteToAdminRequiresStepUp(t *testing.T) {
+	router, _ := setupRoleStepUpRouter(t)
+
+	rec := doJSON(t, router, http.MethodPut, "/api/v1/admin/users/1", map[string]any{"role": "admin"})
+	require.Equal(t, http.StatusUnauthorized, rec.Code)
+}
+
+func TestUpdateUserKeepAdminRoleSkipsStepUp(t *testing.T) {
+	router, _ := setupRoleStepUpRouter(t)
+
+	rec := doJSON(t, router, http.MethodPut, "/api/v1/admin/users/2", map[string]any{"role": "admin"})
+	require.Equal(t, http.StatusOK, rec.Code)
+}
+
+func TestUpdateUserRegularRoleSkipsStepUp(t *testing.T) {
+	router, _ := setupRoleStepUpRouter(t)
+
+	rec := doJSON(t, router, http.MethodPut, "/api/v1/admin/users/1", map[string]any{"role": "user", "email": "u@example.com"})
+	require.Equal(t, http.StatusOK, rec.Code)
+}
+
+func TestCreateAdminUserRequiresStepUp(t *testing.T) {
+	router, _ := setupRoleStepUpRouter(t)
+
+	rec := doJSON(t, router, http.MethodPost, "/api/v1/admin/users", map[string]any{
+		"email": "new-admin@example.com", "password": "pass123", "role": "admin",
+	})
+	require.Equal(t, http.StatusUnauthorized, rec.Code)
+}
+
+func TestCreateRegularUserSkipsStepUp(t *testing.T) {
+	router, _ := setupRoleStepUpRouter(t)
+
+	rec := doJSON(t, router, http.MethodPost, "/api/v1/admin/users", map[string]any{
+		"email": "new-user@example.com", "password": "pass123", "role": "user",
+	})
+	require.Equal(t, http.StatusOK, rec.Code)
+}

backend/internal/handler/auth_handler.go

@@ -107,7 +107,7 @@ func (h *AuthHandler) respondWithTokenPair(c *gin.Context, user *service.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(user)
+		token, tokenErr := h.authService.GenerateToken(c.Request.Context(), user)
 		if tokenErr != nil {
 			response.InternalError(c, "Failed to generate token")
 			return

backend/internal/handler/auth_linuxdo_oauth_test.go

@@ -282,7 +282,7 @@ func TestLinuxDoOAuthBindStartAcceptsAccessTokenCookie(t *testing.T) {
 		Save(context.Background())
 	require.NoError(t, err)
 
-	token, err := handler.authService.GenerateToken(&service.User{
+	token, err := handler.authService.GenerateToken(context.Background(), &service.User{
 		ID:           user.ID,
 		Email:        user.Email,
 		Username:     user.Username,

backend/internal/handler/auth_oauth_pending_flow_test.go

@@ -3360,6 +3360,14 @@ func (s *oauthPendingFlowTotpCacheStub) ClearVerifyAttempts(_ context.Context, u
 	return nil
 }
 
+func (s *oauthPendingFlowTotpCacheStub) SetStepUpGrant(_ context.Context, _ int64, _ string, _ time.Duration) error {
+	return nil
+}
+
+func (s *oauthPendingFlowTotpCacheStub) HasStepUpGrant(_ context.Context, _ int64, _ string) (bool, error) {
+	return false, nil
+}
+
 type oauthPendingFlowTotpEncryptorStub struct{}
 
 func (oauthPendingFlowTotpEncryptorStub) Encrypt(plaintext string) (string, error) {

backend/internal/handler/available_channel_handler.go

@@ -72,6 +72,7 @@ type userSupportedModelPricing struct {
 	OutputPrice      *float64                 `json:"output_price"`
 	CacheWritePrice  *float64                 `json:"cache_write_price"`
 	CacheReadPrice   *float64                 `json:"cache_read_price"`
+	ImageInputPrice  *float64                 `json:"image_input_price"`
 	ImageOutputPrice *float64                 `json:"image_output_price"`
 	PerRequestPrice  *float64                 `json:"per_request_price"`
 	Intervals        []userPricingIntervalDTO `json:"intervals"`
@@ -284,6 +285,7 @@ func toUserPricing(p *service.ChannelModelPricing) *userSupportedModelPricing {
 		OutputPrice:      p.OutputPrice,
 		CacheWritePrice:  p.CacheWritePrice,
 		CacheReadPrice:   p.CacheReadPrice,
+		ImageInputPrice:  p.ImageInputPrice,
 		ImageOutputPrice: p.ImageOutputPrice,
 		PerRequestPrice:  p.PerRequestPrice,
 		Intervals:        intervals,

backend/internal/handler/chat_completions_image_model_test.go

@@ -0,0 +1,96 @@
+package handler
+
+import (
+	"bytes"
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	middleware "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"
+)
+
+func TestChatCompletionsRejectsGPTImageModelsBeforeScheduling(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+
+	for _, model := range []string{"gpt-image-1", "gpt-image-1.5", "gpt-image-2"} {
+		for _, tc := range []struct {
+			name string
+			call func(*gin.Context)
+		}{
+			{
+				name: "gateway",
+				call: (&GatewayHandler{}).ChatCompletions,
+			},
+			{
+				name: "openai_gateway",
+				call: newOpenAIImageChatRejectionHandler(t).ChatCompletions,
+			},
+		} {
+			t.Run(tc.name+"/"+model, func(t *testing.T) {
+				recorder := httptest.NewRecorder()
+				c, _ := gin.CreateTestContext(recorder)
+				body := []byte(`{"model":"` + model + `","messages":[{"role":"user","content":"draw"}]}`)
+				c.Request = httptest.NewRequest(http.MethodPost, "/v1/chat/completions", bytes.NewReader(body))
+				setImageChatTestAuth(c)
+
+				tc.call(c)
+
+				require.Equal(t, http.StatusBadRequest, recorder.Code)
+				require.Equal(t, "invalid_request_error", gjson.Get(recorder.Body.String(), "error.type").String())
+				require.Contains(t, gjson.Get(recorder.Body.String(), "error.message").String(), "Chat Completions")
+				_, selected := c.Get(opsAccountIDKey)
+				require.False(t, selected, "rejection must happen before account selection")
+			})
+		}
+	}
+}
+
+func TestOpenAIChatCompletionsImageModelRejectionDoesNotAcquireConcurrency(t *testing.T) {
+	var acquireCalls atomic.Int64
+	cache := &concurrencyCacheMock{
+		acquireUserSlotFn: func(context.Context, int64, int, string) (bool, error) {
+			acquireCalls.Add(1)
+			return true, nil
+		},
+	}
+	h := newOpenAIImageChatRejectionHandlerWithCache(t, cache)
+	recorder := httptest.NewRecorder()
+	c, _ := gin.CreateTestContext(recorder)
+	c.Request = httptest.NewRequest(http.MethodPost, "/v1/chat/completions", bytes.NewBufferString(
+		`{"model":"gpt-image-2","messages":[{"role":"user","content":"draw"}]}`,
+	))
+	setImageChatTestAuth(c)
+
+	h.ChatCompletions(c)
+
+	require.Equal(t, http.StatusBadRequest, recorder.Code)
+	require.Zero(t, acquireCalls.Load(), "rejection must happen before user/account concurrency and scheduling")
+}
+
+func newOpenAIImageChatRejectionHandler(t *testing.T) *OpenAIGatewayHandler {
+	t.Helper()
+	return newOpenAIImageChatRejectionHandlerWithCache(t, &concurrencyCacheMock{})
+}
+
+func newOpenAIImageChatRejectionHandlerWithCache(t *testing.T, cache *concurrencyCacheMock) *OpenAIGatewayHandler {
+	t.Helper()
+	return &OpenAIGatewayHandler{
+		gatewayService:      &service.OpenAIGatewayService{},
+		billingCacheService: &service.BillingCacheService{},
+		apiKeyService:       &service.APIKeyService{},
+		concurrencyHelper:   NewConcurrencyHelper(service.NewConcurrencyService(cache), SSEPingFormatNone, time.Second),
+	}
+}
+
+func setImageChatTestAuth(c *gin.Context) {
+	apiKey := &service.APIKey{ID: 4348, UserID: 4348, User: &service.User{ID: 4348}}
+	c.Set(string(middleware.ContextKeyAPIKey), apiKey)
+	c.Set(string(middleware.ContextKeyUser), middleware.AuthSubject{UserID: apiKey.UserID, Concurrency: 1})
+}

backend/internal/handler/dto/credentials_redact_test.go

@@ -20,6 +20,7 @@ func TestRedactCredentials_StripsSensitiveKeysAndReportsStatus(t *testing.T) {
 		"aws_secret_access_key": "aws-secret",
 		"service_account_json":  map[string]any{"private_key": "..."},
 		"private_key":           "raw-key",
+		"agent_private_key":     "agent-key-secret",
 		// 非敏感
 		"base_url":      "https://api.example.com",
 		"model_mapping": map[string]any{"foo": "bar"},
@@ -35,6 +36,7 @@ func TestRedactCredentials_StripsSensitiveKeysAndReportsStatus(t *testing.T) {
 	require.NotContains(t, out, "aws_secret_access_key")
 	require.NotContains(t, out, "service_account_json")
 	require.NotContains(t, out, "private_key")
+	require.NotContains(t, out, "agent_private_key")
 
 	require.Equal(t, "https://api.example.com", out["base_url"])
 	require.Equal(t, map[string]any{"foo": "bar"}, out["model_mapping"])
@@ -47,6 +49,7 @@ func TestRedactCredentials_StripsSensitiveKeysAndReportsStatus(t *testing.T) {
 	require.True(t, status["has_aws_secret_access_key"])
 	require.True(t, status["has_service_account_json"])
 	require.True(t, status["has_private_key"])
+	require.True(t, status["has_agent_private_key"])
 
 	// 状态 map 不应携带非敏感键的 has_*
 	require.NotContains(t, status, "has_base_url")
@@ -84,6 +87,7 @@ func TestRedactCredentials_AllKnownSensitiveKeys(t *testing.T) {
 		"api_key", "session_key", "cookie",
 		"aws_secret_access_key", "aws_session_token",
 		"service_account_json", "service_account", "private_key",
+		"agent_private_key",
 	}
 	in := make(map[string]any, len(keys))
 	for _, k := range keys {

backend/internal/handler/dto/mappers.go

@@ -199,6 +199,7 @@ func groupFromServiceBase(g *service.Group) Group {
 		VideoPrice480P:                  g.VideoPrice480P,
 		VideoPrice720P:                  g.VideoPrice720P,
 		VideoPrice1080P:                 g.VideoPrice1080P,
+		WebSearchPricePerCall:           g.WebSearchPricePerCall,
 		ClaudeCodeOnly:                  g.ClaudeCodeOnly,
 		FallbackGroupID:                 g.FallbackGroupID,
 		FallbackGroupIDOnInvalidRequest: g.FallbackGroupIDOnInvalidRequest,
@@ -598,54 +599,57 @@ func usageLogFromServiceUser(l *service.UsageLog) UsageLog {
 		requestedModel = l.Model
 	}
 	return UsageLog{
-		ID:                    l.ID,
-		UserID:                l.UserID,
-		APIKeyID:              l.APIKeyID,
-		AccountID:             l.AccountID,
-		RequestID:             l.RequestID,
-		Model:                 requestedModel,
-		ServiceTier:           l.ServiceTier,
-		ReasoningEffort:       l.ReasoningEffort,
-		InboundEndpoint:       l.InboundEndpoint,
-		GroupID:               l.GroupID,
-		SubscriptionID:        l.SubscriptionID,
-		InputTokens:           l.InputTokens,
-		OutputTokens:          l.OutputTokens,
-		CacheCreationTokens:   l.CacheCreationTokens,
-		CacheReadTokens:       l.CacheReadTokens,
-		CacheCreation5mTokens: l.CacheCreation5mTokens,
-		CacheCreation1hTokens: l.CacheCreation1hTokens,
-		InputCost:             l.InputCost,
-		OutputCost:            l.OutputCost,
-		CacheCreationCost:     l.CacheCreationCost,
-		CacheReadCost:         l.CacheReadCost,
-		TotalCost:             l.TotalCost,
-		ActualCost:            l.ActualCost,
-		RateMultiplier:        l.RateMultiplier,
-		BillingType:           l.BillingType,
-		RequestType:           requestType.String(),
-		Stream:                stream,
-		OpenAIWSMode:          openAIWSMode,
-		DurationMs:            l.DurationMs,
-		FirstTokenMs:          l.FirstTokenMs,
-		ImageCount:            l.ImageCount,
-		ImageSize:             l.ImageSize,
-		ImageInputSize:        l.ImageInputSize,
-		ImageOutputSize:       l.ImageOutputSize,
-		ImageOutputTokens:     l.ImageOutputTokens,
-		ImageOutputCost:       l.ImageOutputCost,
-		ImageSizeSource:       l.ImageSizeSource,
-		ImageSizeBreakdown:    l.ImageSizeBreakdown,
-		MediaType:             l.MediaType,
-		UserAgent:             l.UserAgent,
-		IPAddress:             l.IPAddress,
-		CacheTTLOverridden:    l.CacheTTLOverridden,
-		BillingMode:           l.BillingMode,
-		CreatedAt:             l.CreatedAt,
-		User:                  UserFromServiceShallow(l.User),
-		APIKey:                APIKeyFromService(l.APIKey),
-		Group:                 GroupFromServiceShallow(l.Group),
-		Subscription:          UserSubscriptionFromService(l.Subscription),
+		ID:                        l.ID,
+		UserID:                    l.UserID,
+		APIKeyID:                  l.APIKeyID,
+		AccountID:                 l.AccountID,
+		RequestID:                 l.RequestID,
+		Model:                     requestedModel,
+		ServiceTier:               l.ServiceTier,
+		ReasoningEffort:           l.ReasoningEffort,
+		InboundEndpoint:           l.InboundEndpoint,
+		GroupID:                   l.GroupID,
+		SubscriptionID:            l.SubscriptionID,
+		InputTokens:               l.InputTokens,
+		OutputTokens:              l.OutputTokens,
+		CacheCreationTokens:       l.CacheCreationTokens,
+		CacheReadTokens:           l.CacheReadTokens,
+		CacheCreation5mTokens:     l.CacheCreation5mTokens,
+		CacheCreation1hTokens:     l.CacheCreation1hTokens,
+		InputCost:                 l.InputCost,
+		OutputCost:                l.OutputCost,
+		CacheCreationCost:         l.CacheCreationCost,
+		CacheReadCost:             l.CacheReadCost,
+		TotalCost:                 l.TotalCost,
+		ActualCost:                l.ActualCost,
+		RateMultiplier:            l.RateMultiplier,
+		LongContextBillingApplied: l.LongContextBillingApplied,
+		BillingType:               l.BillingType,
+		RequestType:               requestType.String(),
+		Stream:                    stream,
+		OpenAIWSMode:              openAIWSMode,
+		DurationMs:                l.DurationMs,
+		FirstTokenMs:              l.FirstTokenMs,
+		ImageCount:                l.ImageCount,
+		ImageSize:                 l.ImageSize,
+		ImageInputSize:            l.ImageInputSize,
+		ImageOutputSize:           l.ImageOutputSize,
+		ImageInputTokens:          l.ImageInputTokens,
+		ImageInputCost:            l.ImageInputCost,
+		ImageOutputTokens:         l.ImageOutputTokens,
+		ImageOutputCost:           l.ImageOutputCost,
+		ImageSizeSource:           l.ImageSizeSource,
+		ImageSizeBreakdown:        l.ImageSizeBreakdown,
+		MediaType:                 l.MediaType,
+		UserAgent:                 l.UserAgent,
+		IPAddress:                 l.IPAddress,
+		CacheTTLOverridden:        l.CacheTTLOverridden,
+		BillingMode:               l.BillingMode,
+		CreatedAt:                 l.CreatedAt,
+		User:                      UserFromServiceShallow(l.User),
+		APIKey:                    APIKeyFromService(l.APIKey),
+		Group:                     GroupFromServiceShallow(l.Group),
+		Subscription:              UserSubscriptionFromService(l.Subscription),
 	}
 }
 

backend/internal/handler/dto/settings.go

@@ -36,6 +36,8 @@ 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
+	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"`
@@ -148,6 +150,7 @@ type SystemSettings struct {
 	AffiliateRebateFreezeHours   int                          `json:"affiliate_rebate_freeze_hours"`
 	AffiliateRebateDurationDays  int                          `json:"affiliate_rebate_duration_days"`
 	AffiliateRebatePerInviteeCap float64                      `json:"affiliate_rebate_per_invitee_cap"`
+	AdminRechargeRebateEnabled   bool                         `json:"affiliate_admin_recharge_enabled"`
 	DefaultUserRPMLimit          int                          `json:"default_user_rpm_limit"`
 	DefaultSubscriptions         []DefaultSubscriptionSetting `json:"default_subscriptions"`
 
@@ -208,29 +211,33 @@ type SystemSettings struct {
 	PaymentVisibleMethodWxpayEnabled  bool   `json:"payment_visible_method_wxpay_enabled"`
 
 	// OpenAI account scheduling
-	OpenAIAdvancedSchedulerEnabled                         bool   `json:"openai_advanced_scheduler_enabled"`
-	OpenAIAdvancedSchedulerStickyWeightedEnabled           bool   `json:"openai_advanced_scheduler_sticky_weighted_enabled"`
-	OpenAIAdvancedSchedulerSubscriptionPriorityEnabled     bool   `json:"openai_advanced_scheduler_subscription_priority_enabled"`
-	OpenAIAdvancedSchedulerLBTopK                          string `json:"openai_advanced_scheduler_lb_top_k"`
-	OpenAIAdvancedSchedulerWeightPriority                  string `json:"openai_advanced_scheduler_weight_priority"`
-	OpenAIAdvancedSchedulerWeightLoad                      string `json:"openai_advanced_scheduler_weight_load"`
-	OpenAIAdvancedSchedulerWeightQueue                     string `json:"openai_advanced_scheduler_weight_queue"`
-	OpenAIAdvancedSchedulerWeightErrorRate                 string `json:"openai_advanced_scheduler_weight_error_rate"`
-	OpenAIAdvancedSchedulerWeightTTFT                      string `json:"openai_advanced_scheduler_weight_ttft"`
-	OpenAIAdvancedSchedulerWeightReset                     string `json:"openai_advanced_scheduler_weight_reset"`
-	OpenAIAdvancedSchedulerWeightQuotaHeadroom             string `json:"openai_advanced_scheduler_weight_quota_headroom"`
-	OpenAIAdvancedSchedulerWeightPreviousResponse          string `json:"openai_advanced_scheduler_weight_previous_response"`
-	OpenAIAdvancedSchedulerWeightSessionSticky             string `json:"openai_advanced_scheduler_weight_session_sticky"`
-	OpenAIAdvancedSchedulerEffectiveLBTopK                 string `json:"openai_advanced_scheduler_effective_lb_top_k"`
-	OpenAIAdvancedSchedulerEffectiveWeightPriority         string `json:"openai_advanced_scheduler_effective_weight_priority"`
-	OpenAIAdvancedSchedulerEffectiveWeightLoad             string `json:"openai_advanced_scheduler_effective_weight_load"`
-	OpenAIAdvancedSchedulerEffectiveWeightQueue            string `json:"openai_advanced_scheduler_effective_weight_queue"`
-	OpenAIAdvancedSchedulerEffectiveWeightErrorRate        string `json:"openai_advanced_scheduler_effective_weight_error_rate"`
-	OpenAIAdvancedSchedulerEffectiveWeightTTFT             string `json:"openai_advanced_scheduler_effective_weight_ttft"`
-	OpenAIAdvancedSchedulerEffectiveWeightReset            string `json:"openai_advanced_scheduler_effective_weight_reset"`
-	OpenAIAdvancedSchedulerEffectiveWeightQuotaHeadroom    string `json:"openai_advanced_scheduler_effective_weight_quota_headroom"`
-	OpenAIAdvancedSchedulerEffectiveWeightPreviousResponse string `json:"openai_advanced_scheduler_effective_weight_previous_response"`
-	OpenAIAdvancedSchedulerEffectiveWeightSessionSticky    string `json:"openai_advanced_scheduler_effective_weight_session_sticky"`
+	OpenAILowUpstreamRatePriorityEnabled                   bool    `json:"openai_low_upstream_rate_priority_enabled"`
+	OpenAIOAuthSchedulingRateMultiplier                    float64 `json:"openai_oauth_scheduling_rate_multiplier"`
+	OpenAIAdvancedSchedulerEnabled                         bool    `json:"openai_advanced_scheduler_enabled"`
+	OpenAIAdvancedSchedulerStickyWeightedEnabled           bool    `json:"openai_advanced_scheduler_sticky_weighted_enabled"`
+	OpenAIAdvancedSchedulerSubscriptionPriorityEnabled     bool    `json:"openai_advanced_scheduler_subscription_priority_enabled"`
+	OpenAIAdvancedSchedulerLBTopK                          string  `json:"openai_advanced_scheduler_lb_top_k"`
+	OpenAIAdvancedSchedulerWeightPriority                  string  `json:"openai_advanced_scheduler_weight_priority"`
+	OpenAIAdvancedSchedulerWeightLoad                      string  `json:"openai_advanced_scheduler_weight_load"`
+	OpenAIAdvancedSchedulerWeightQueue                     string  `json:"openai_advanced_scheduler_weight_queue"`
+	OpenAIAdvancedSchedulerWeightErrorRate                 string  `json:"openai_advanced_scheduler_weight_error_rate"`
+	OpenAIAdvancedSchedulerWeightTTFT                      string  `json:"openai_advanced_scheduler_weight_ttft"`
+	OpenAIAdvancedSchedulerWeightReset                     string  `json:"openai_advanced_scheduler_weight_reset"`
+	OpenAIAdvancedSchedulerWeightQuotaHeadroom             string  `json:"openai_advanced_scheduler_weight_quota_headroom"`
+	OpenAIAdvancedSchedulerWeightUpstreamCost              string  `json:"openai_advanced_scheduler_weight_upstream_cost"`
+	OpenAIAdvancedSchedulerWeightPreviousResponse          string  `json:"openai_advanced_scheduler_weight_previous_response"`
+	OpenAIAdvancedSchedulerWeightSessionSticky             string  `json:"openai_advanced_scheduler_weight_session_sticky"`
+	OpenAIAdvancedSchedulerEffectiveLBTopK                 string  `json:"openai_advanced_scheduler_effective_lb_top_k"`
+	OpenAIAdvancedSchedulerEffectiveWeightPriority         string  `json:"openai_advanced_scheduler_effective_weight_priority"`
+	OpenAIAdvancedSchedulerEffectiveWeightLoad             string  `json:"openai_advanced_scheduler_effective_weight_load"`
+	OpenAIAdvancedSchedulerEffectiveWeightQueue            string  `json:"openai_advanced_scheduler_effective_weight_queue"`
+	OpenAIAdvancedSchedulerEffectiveWeightErrorRate        string  `json:"openai_advanced_scheduler_effective_weight_error_rate"`
+	OpenAIAdvancedSchedulerEffectiveWeightTTFT             string  `json:"openai_advanced_scheduler_effective_weight_ttft"`
+	OpenAIAdvancedSchedulerEffectiveWeightReset            string  `json:"openai_advanced_scheduler_effective_weight_reset"`
+	OpenAIAdvancedSchedulerEffectiveWeightQuotaHeadroom    string  `json:"openai_advanced_scheduler_effective_weight_quota_headroom"`
+	OpenAIAdvancedSchedulerEffectiveWeightUpstreamCost     string  `json:"openai_advanced_scheduler_effective_weight_upstream_cost"`
+	OpenAIAdvancedSchedulerEffectiveWeightPreviousResponse string  `json:"openai_advanced_scheduler_effective_weight_previous_response"`
+	OpenAIAdvancedSchedulerEffectiveWeightSessionSticky    string  `json:"openai_advanced_scheduler_effective_weight_session_sticky"`
 
 	// Payment configuration
 	PaymentEnabled                   bool     `json:"payment_enabled"`

backend/internal/handler/dto/types.go

@@ -120,6 +120,8 @@ type Group struct {
 	VideoPrice480P     *float64 `json:"video_price_480p"`
 	VideoPrice720P     *float64 `json:"video_price_720p"`
 	VideoPrice1080P    *float64 `json:"video_price_1080p"`
+	// Codex alpha/search 网页搜索单次价格（USD/次）；null 表示使用默认价 0.01
+	WebSearchPricePerCall *float64 `json:"web_search_price_per_call"`
 
 	// Claude Code 客户端限制
 	ClaudeCodeOnly  bool   `json:"claude_code_only"`
@@ -480,13 +482,14 @@ type UsageLog struct {
 	CacheCreation5mTokens int `json:"cache_creation_5m_tokens"`
 	CacheCreation1hTokens int `json:"cache_creation_1h_tokens"`
 
-	InputCost         float64 `json:"input_cost"`
-	OutputCost        float64 `json:"output_cost"`
-	CacheCreationCost float64 `json:"cache_creation_cost"`
-	CacheReadCost     float64 `json:"cache_read_cost"`
-	TotalCost         float64 `json:"total_cost"`
-	ActualCost        float64 `json:"actual_cost"`
-	RateMultiplier    float64 `json:"rate_multiplier"`
+	InputCost                 float64 `json:"input_cost"`
+	OutputCost                float64 `json:"output_cost"`
+	CacheCreationCost         float64 `json:"cache_creation_cost"`
+	CacheReadCost             float64 `json:"cache_read_cost"`
+	TotalCost                 float64 `json:"total_cost"`
+	ActualCost                float64 `json:"actual_cost"`
+	RateMultiplier            float64 `json:"rate_multiplier"`
+	LongContextBillingApplied bool    `json:"long_context_billing_applied"`
 
 	BillingType  int8   `json:"billing_type"`
 	RequestType  string `json:"request_type"`
@@ -500,6 +503,8 @@ type UsageLog struct {
 	ImageSize          *string        `json:"image_size"`
 	ImageInputSize     *string        `json:"image_input_size"`
 	ImageOutputSize    *string        `json:"image_output_size"`
+	ImageInputTokens   int            `json:"image_input_tokens"`
+	ImageInputCost     float64        `json:"image_input_cost"`
 	ImageOutputTokens  int            `json:"image_output_tokens"`
 	ImageOutputCost    float64        `json:"image_output_cost"`
 	ImageSizeSource    *string        `json:"image_size_source"`

backend/internal/handler/endpoint.go

@@ -18,11 +18,15 @@ const (
 	EndpointMessages          = "/v1/messages"
 	EndpointChatCompletions   = "/v1/chat/completions"
 	EndpointEmbeddings        = "/v1/embeddings"
+	EndpointAlphaSearch       = "/v1/alpha/search"
 	EndpointResponses         = "/v1/responses"
 	EndpointResponsesCompact  = "/v1/responses/compact"
 	EndpointImagesGenerations = "/v1/images/generations"
 	EndpointImagesEdits       = "/v1/images/edits"
+	EndpointImageTasks        = "/v1/images/tasks"
 	EndpointVideosGenerations = "/v1/videos/generations"
+	EndpointVideosEdits       = "/v1/videos/edits"
+	EndpointVideosExtensions  = "/v1/videos/extensions"
 	EndpointVideos            = "/v1/videos"
 	EndpointGeminiModels      = "/v1beta/models"
 )
@@ -75,6 +79,8 @@ func NormalizeInboundEndpoint(path string) string {
 	switch {
 	case strings.Contains(path, EndpointEmbeddings):
 		return EndpointEmbeddings
+	case strings.Contains(path, EndpointAlphaSearch) || isBareOrSubpathOf(strings.TrimRight(path, "/"), "/alpha/search") || isBareOrSubpathOf(strings.TrimRight(path, "/"), "/backend-api/codex/alpha/search"):
+		return EndpointAlphaSearch
 	case strings.Contains(path, EndpointChatCompletions):
 		return EndpointChatCompletions
 	case strings.Contains(path, EndpointMessages):
@@ -83,8 +89,14 @@ func NormalizeInboundEndpoint(path string) string {
 		return EndpointImagesGenerations
 	case strings.Contains(path, EndpointImagesEdits) || strings.Contains(path, "/images/edits"):
 		return EndpointImagesEdits
+	case strings.Contains(path, EndpointImageTasks) || strings.Contains(path, "/images/tasks/"):
+		return EndpointImageTasks
 	case strings.Contains(path, EndpointVideosGenerations) || strings.Contains(path, "/videos/generations"):
 		return EndpointVideosGenerations
+	case strings.Contains(path, EndpointVideosEdits) || strings.Contains(path, "/videos/edits"):
+		return EndpointVideosEdits
+	case strings.Contains(path, EndpointVideosExtensions) || strings.Contains(path, "/videos/extensions"):
+		return EndpointVideosExtensions
 	case strings.Contains(path, EndpointVideos) || strings.Contains(path, "/videos/"):
 		return EndpointVideos
 	case strings.Contains(path, EndpointResponsesCompact) || isResponsesCompactAliasPath(path):
@@ -155,8 +167,11 @@ func isBareOrSubpathOf(path, root string) bool {
 // account platform and the normalized inbound endpoint.
 //
 // Platform-specific rules:
-//   - OpenAI always forwards to /v1/responses (with optional subpath
-//     such as /v1/responses/compact preserved from the raw URL).
+//   - OpenAI and Grok text compatibility routes forward to /v1/responses
+//     (with optional subpath such as /v1/responses/compact preserved from
+//     the raw URL); native endpoints such as embeddings and alpha search
+//     retain their paths. Grok raw Chat requests override this through the
+//     forwarding result consumed by resolveOpenAIUpstreamEndpoint.
 //   - Anthropic  → /v1/messages
 //   - Gemini     → /v1beta/models
 //   - Antigravity → /v1/messages (Claude) or gemini (Gemini)
@@ -167,7 +182,7 @@ func DeriveUpstreamEndpoint(inbound, rawRequestPath, platform string) string {
 
 	switch platform {
 	case service.PlatformOpenAI, service.PlatformGrok:
-		if inbound == EndpointEmbeddings || inbound == EndpointImagesGenerations || inbound == EndpointImagesEdits || inbound == EndpointVideosGenerations || inbound == EndpointVideos {
+		if inbound == EndpointEmbeddings || inbound == EndpointAlphaSearch || inbound == EndpointImagesGenerations || inbound == EndpointImagesEdits || inbound == EndpointVideosGenerations || inbound == EndpointVideosEdits || inbound == EndpointVideosExtensions || inbound == EndpointVideos {
 			return inbound
 		}
 		// OpenAI forwards everything to the Responses API.

backend/internal/handler/endpoint_test.go

@@ -25,11 +25,13 @@ func TestNormalizeInboundEndpoint(t *testing.T) {
 		{"/v1/messages", EndpointMessages},
 		{"/v1/chat/completions", EndpointChatCompletions},
 		{"/v1/embeddings", EndpointEmbeddings},
+		{"/v1/alpha/search", EndpointAlphaSearch},
 		{"/v1/responses", EndpointResponses},
 		{"/v1/responses/compact", EndpointResponsesCompact},
 		{"/v1/responses/compact/detail", EndpointResponsesCompact},
 		{"/v1/images/generations", EndpointImagesGenerations},
 		{"/v1/images/edits", EndpointImagesEdits},
+		{"/v1/images/tasks/imgtask_123", EndpointImageTasks},
 		{"/v1/videos/generations", EndpointVideosGenerations},
 		{"/v1/videos/req_123", EndpointVideos},
 		{"/v1beta/models", EndpointGeminiModels},
@@ -50,11 +52,14 @@ func TestNormalizeInboundEndpoint(t *testing.T) {
 		{"/responses", EndpointResponses},
 		{"/responses/compact", EndpointResponsesCompact},
 		{"/responses/compact/detail", EndpointResponsesCompact},
+		{"/alpha/search", EndpointAlphaSearch},
+		{"/images/tasks/imgtask_123", EndpointImageTasks},
 
 		// Bare Codex direct alias route — root vs. compact.
 		{"/backend-api/codex/responses", EndpointResponses},
 		{"/backend-api/codex/responses/compact", EndpointResponsesCompact},
 		{"/backend-api/codex/responses/compact/detail", EndpointResponsesCompact},
+		{"/backend-api/codex/alpha/search", EndpointAlphaSearch},
 
 		// Must NOT generalize to arbitrary paths merely ending in
 		// "/responses" (or "/responses/compact") that are unrelated to
@@ -119,8 +124,11 @@ func TestDeriveUpstreamEndpoint(t *testing.T) {
 		{"openai from messages", EndpointMessages, "/v1/messages", service.PlatformOpenAI, EndpointResponses},
 		{"openai from completions", EndpointChatCompletions, "/v1/chat/completions", service.PlatformOpenAI, EndpointResponses},
 		{"openai embeddings", EndpointEmbeddings, "/v1/embeddings", service.PlatformOpenAI, EndpointEmbeddings},
+		{"openai alpha search", EndpointAlphaSearch, "/backend-api/codex/alpha/search", service.PlatformOpenAI, EndpointAlphaSearch},
 		{"openai image generations", EndpointImagesGenerations, "/v1/images/generations", service.PlatformOpenAI, EndpointImagesGenerations},
 		{"openai image edits", EndpointImagesEdits, "/openai/v1/images/edits", service.PlatformOpenAI, EndpointImagesEdits},
+		{"grok chat defaults to responses without runtime result", EndpointChatCompletions, "/v1/chat/completions", service.PlatformGrok, EndpointResponses},
+		{"grok responses", EndpointResponses, "/v1/responses", service.PlatformGrok, EndpointResponses},
 		{"grok video generations", EndpointVideosGenerations, "/v1/videos/generations", service.PlatformGrok, EndpointVideosGenerations},
 		{"grok video status", EndpointVideos, "/videos/req_123", service.PlatformGrok, EndpointVideos},
 
@@ -138,6 +146,59 @@ func TestDeriveUpstreamEndpoint(t *testing.T) {
 	}
 }
 
+func TestResolveOpenAIUpstreamEndpointPrefersForwardResult(t *testing.T) {
+	tests := []struct {
+		name            string
+		account         *service.Account
+		result          *service.OpenAIForwardResult
+		runtimeEndpoint string
+		want            string
+	}{
+		{
+			name:            "grok raw chat result overrides stale context",
+			account:         &service.Account{Platform: service.PlatformGrok, Type: service.AccountTypeOAuth},
+			result:          &service.OpenAIForwardResult{UpstreamEndpoint: EndpointChatCompletions},
+			runtimeEndpoint: EndpointResponses,
+			want:            EndpointChatCompletions,
+		},
+		{
+			name:    "grok chat bridged to responses",
+			account: &service.Account{Platform: service.PlatformGrok, Type: service.AccountTypeOAuth},
+			result:  &service.OpenAIForwardResult{UpstreamEndpoint: EndpointResponses},
+			want:    EndpointResponses,
+		},
+		{
+			name:    "grok empty result keeps responses default",
+			account: &service.Account{Platform: service.PlatformGrok, Type: service.AccountTypeOAuth},
+			result:  &service.OpenAIForwardResult{},
+			want:    EndpointResponses,
+		},
+		{
+			name:            "grok raw error uses runtime endpoint",
+			account:         &service.Account{Platform: service.PlatformGrok, Type: service.AccountTypeOAuth},
+			runtimeEndpoint: EndpointChatCompletions,
+			want:            EndpointChatCompletions,
+		},
+		{
+			name:    "openai behavior remains responses",
+			account: &service.Account{Platform: service.PlatformOpenAI, Type: service.AccountTypeOAuth},
+			result:  &service.OpenAIForwardResult{},
+			want:    EndpointResponses,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			rec := httptest.NewRecorder()
+			c, _ := gin.CreateTestContext(rec)
+			c.Request = httptest.NewRequest(http.MethodPost, EndpointChatCompletions, nil)
+			c.Set(ctxKeyInboundEndpoint, EndpointChatCompletions)
+			service.SetActualOpenAIUpstreamEndpoint(c, tt.runtimeEndpoint)
+			require.Equal(t, tt.want, resolveOpenAIUpstreamEndpoint(c, tt.account, tt.result))
+		})
+	}
+}
+
 // ──────────────────────────────────────────────────────────
 // responsesSubpathSuffix
 // ──────────────────────────────────────────────────────────

backend/internal/handler/failover_loop.go

@@ -5,6 +5,8 @@ import (
 	"net/http"
 	"time"
 
+	"github.com/gin-gonic/gin"
+
 	"github.com/Wei-Shaw/sub2api/internal/pkg/logger"
 	"github.com/Wei-Shaw/sub2api/internal/service"
 	"go.uber.org/zap"
@@ -29,7 +31,8 @@ const (
 )
 
 const (
-	// maxSameAccountRetries 同账号重试次数上限（针对 RetryableOnSameAccount 错误）
+	// maxSameAccountRetries 同账号重试次数默认上限（针对 RetryableOnSameAccount 错误）。
+	// 生产调用方通常传入账号级配置 account.GetPoolModeRetryCount()，该常量仅作兜底/测试默认值。
 	maxSameAccountRetries = 3
 	// sameAccountRetryDelay 同账号重试间隔
 	sameAccountRetryDelay = 500 * time.Millisecond
@@ -67,23 +70,33 @@ func (s *FailoverState) HandleFailoverError(
 	gatewayService TempUnscheduler,
 	accountID int64,
 	platform string,
+	retryLimit int,
 	failoverErr *service.UpstreamFailoverError,
 ) FailoverAction {
+	// 客户端已断开：failover 只会用已取消的 context 重新选号并必然失败，
+	// 不应再被当成账号耗尽处理（误报 502）。
+	if ctx != nil && ctx.Err() != nil {
+		return FailoverCanceled
+	}
 	s.LastFailoverErr = failoverErr
+	if failoverErr == nil || !failoverErr.ShouldRetryNextAccount() {
+		return FailoverExhausted
+	}
 
 	// 缓存计费判断
 	if needForceCacheBilling(s.hasBoundSession, failoverErr) {
 		s.ForceCacheBilling = true
 	}
 
-	// 同账号重试：对 RetryableOnSameAccount 的临时性错误，先在同一账号上重试
-	if failoverErr.RetryableOnSameAccount && s.SameAccountRetryCount[accountID] < maxSameAccountRetries {
+	// 同账号重试：对 RetryableOnSameAccount 的临时性错误，先在同一账号上重试。
+	// 重试次数上限 retryLimit 由调用方传入（账号级 pool_mode_retry_count 配置）。
+	if failoverErr.RetryableOnSameAccount && s.SameAccountRetryCount[accountID] < retryLimit {
 		s.SameAccountRetryCount[accountID]++
 		logger.FromContext(ctx).Warn("gateway.failover_same_account_retry",
 			zap.Int64("account_id", accountID),
 			zap.Int("upstream_status", failoverErr.StatusCode),
 			zap.Int("same_account_retry_count", s.SameAccountRetryCount[accountID]),
-			zap.Int("same_account_retry_max", maxSameAccountRetries),
+			zap.Int("same_account_retry_max", retryLimit),
 		)
 		if !sleepWithContext(ctx, sameAccountRetryDelay) {
 			return FailoverCanceled
@@ -132,6 +145,12 @@ func (s *FailoverState) HandleFailoverError(
 // 返回 FailoverExhausted 时，调用方应返回错误响应。
 // 返回 FailoverCanceled 时，调用方应直接 return。
 func (s *FailoverState) HandleSelectionExhausted(ctx context.Context) FailoverAction {
+	// 客户端已断开时选号失败是 context canceled 的必然结果，
+	// 不代表账号耗尽，直接按取消终止。
+	if ctx.Err() != nil {
+		return FailoverCanceled
+	}
+
 	if s.LastFailoverErr != nil &&
 		s.LastFailoverErr.StatusCode == http.StatusServiceUnavailable &&
 		s.SwitchCount <= s.MaxSwitches {
@@ -160,6 +179,28 @@ func needForceCacheBilling(hasBoundSession bool, failoverErr *service.UpstreamFa
 	return hasBoundSession || (failoverErr != nil && failoverErr.ForceCacheBilling)
 }
 
+// failoverClientGone 判断下游客户端是否已断开（请求 context 已取消）。
+// 客户端断开后 failover 必须静默终止：用已取消的 context 重新选号只会得到
+// context.Canceled，并被误报成账号耗尽（通用 502）；上游 detach 的在途请求
+// 照常完成计费，但不再为无人接收的响应启动新的上游尝试。
+// 响应尚未提交时把状态码标记为 499（client closed request），供访问日志归类。
+func failoverClientGone(c *gin.Context) bool {
+	if c == nil || c.Request == nil || c.Request.Context().Err() == nil {
+		return false
+	}
+	// 先停 compact 心跳（接管 ResponseWriter，建立 happens-before），与
+	// handleStreamingAwareError/errorResponse 等终结路径对齐，避免心跳
+	// goroutine 与下面的状态标记并发触碰同一 writer。心跳已提交 200 时
+	// 状态码已固化，不再标 499。
+	if service.StopOpenAICompactSSEKeepaliveCommitted(c) {
+		return true
+	}
+	if !c.Writer.Written() {
+		c.Status(statusClientClosedRequest)
+	}
+	return true
+}
+
 // sleepWithContext 等待指定时长，返回 false 表示 context 已取消。
 func sleepWithContext(ctx context.Context, d time.Duration) bool {
 	if d <= 0 {

backend/internal/handler/failover_loop_test.go

@@ -2,11 +2,14 @@ package handler
 
 import (
 	"context"
+	"net/http"
+	"net/http/httptest"
 	"testing"
 	"time"
 
 	"github.com/Wei-Shaw/sub2api/internal/service"
 
+	"github.com/gin-gonic/gin"
 	"github.com/stretchr/testify/require"
 )
 
@@ -128,12 +131,56 @@ func TestSleepWithContext(t *testing.T) {
 // ---------------------------------------------------------------------------
 
 func TestHandleFailoverError_BasicSwitch(t *testing.T) {
+	t.Run("显式停止不切换账号且旧错误默认仍切换", func(t *testing.T) {
+		mock := &mockTempUnscheduler{}
+		fs := NewFailoverState(3, false)
+		stopErr := &service.UpstreamFailoverError{
+			Stage:             service.GatewayFailureStageAccountAuth,
+			Scope:             service.GatewayFailureScopeProvider,
+			NextAccountAction: service.NextAccountStop,
+		}
+
+		action := fs.HandleFailoverError(context.Background(), mock, 100, service.PlatformGrok, maxSameAccountRetries, stopErr)
+
+		require.Equal(t, FailoverExhausted, action)
+		require.Zero(t, fs.SwitchCount)
+		require.Empty(t, fs.FailedAccountIDs)
+		require.Equal(t, stopErr, fs.LastFailoverErr)
+
+		legacyErr := newTestFailoverErr(http.StatusTooManyRequests, false, false)
+		action = fs.HandleFailoverError(context.Background(), mock, 100, service.PlatformGrok, maxSameAccountRetries, legacyErr)
+
+		require.Equal(t, FailoverContinue, action)
+		require.Equal(t, 1, fs.SwitchCount)
+		require.Contains(t, fs.FailedAccountIDs, int64(100))
+	})
+
+	t.Run("已取消的认证失败不改变切换状态", func(t *testing.T) {
+		ctx, cancel := context.WithCancel(context.Background())
+		cancel()
+		mock := &mockTempUnscheduler{}
+		fs := NewFailoverState(3, false)
+		err := &service.UpstreamFailoverError{
+			Stage:             service.GatewayFailureStageAccountAuth,
+			Scope:             service.GatewayFailureScopeAccount,
+			NextAccountAction: service.NextAccountRetry,
+		}
+
+		action := fs.HandleFailoverError(ctx, mock, 101, service.PlatformGrok, maxSameAccountRetries, err)
+
+		require.Equal(t, FailoverCanceled, action)
+		require.Zero(t, fs.SwitchCount)
+		require.Empty(t, fs.FailedAccountIDs)
+		require.Nil(t, fs.LastFailoverErr)
+		require.Empty(t, mock.calls)
+	})
+
 	t.Run("非重试错误_非Antigravity_直接切换", func(t *testing.T) {
 		mock := &mockTempUnscheduler{}
 		fs := NewFailoverState(3, false)
 		err := newTestFailoverErr(500, false, false)
 
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 1, fs.SwitchCount)
@@ -150,7 +197,7 @@ func TestHandleFailoverError_BasicSwitch(t *testing.T) {
 		err := newTestFailoverErr(500, false, false)
 
 		start := time.Now()
-		action := fs.HandleFailoverError(context.Background(), mock, 100, service.PlatformAntigravity, err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, service.PlatformAntigravity, maxSameAccountRetries, err)
 		elapsed := time.Since(start)
 
 		require.Equal(t, FailoverContinue, action)
@@ -166,7 +213,7 @@ func TestHandleFailoverError_BasicSwitch(t *testing.T) {
 
 		err := newTestFailoverErr(500, false, false)
 		start := time.Now()
-		action := fs.HandleFailoverError(context.Background(), mock, 200, service.PlatformAntigravity, err)
+		action := fs.HandleFailoverError(context.Background(), mock, 200, service.PlatformAntigravity, maxSameAccountRetries, err)
 		elapsed := time.Since(start)
 
 		require.Equal(t, FailoverContinue, action)
@@ -181,19 +228,19 @@ func TestHandleFailoverError_BasicSwitch(t *testing.T) {
 
 		// 第一次切换：0→1
 		err1 := newTestFailoverErr(500, false, false)
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", err1)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err1)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 1, fs.SwitchCount)
 
 		// 第二次切换：1→2
 		err2 := newTestFailoverErr(502, false, false)
-		action = fs.HandleFailoverError(context.Background(), mock, 200, "openai", err2)
+		action = fs.HandleFailoverError(context.Background(), mock, 200, "openai", maxSameAccountRetries, err2)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 2, fs.SwitchCount)
 
 		// 第三次已耗尽：SwitchCount(2) >= MaxSwitches(2)
 		err3 := newTestFailoverErr(503, false, false)
-		action = fs.HandleFailoverError(context.Background(), mock, 300, "openai", err3)
+		action = fs.HandleFailoverError(context.Background(), mock, 300, "openai", maxSameAccountRetries, err3)
 		require.Equal(t, FailoverExhausted, action)
 		require.Equal(t, 2, fs.SwitchCount, "耗尽时不应继续递增")
 
@@ -212,7 +259,7 @@ func TestHandleFailoverError_BasicSwitch(t *testing.T) {
 		fs := NewFailoverState(0, false)
 		err := newTestFailoverErr(500, false, false)
 
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.Equal(t, FailoverExhausted, action)
 		require.Equal(t, 0, fs.SwitchCount)
 		require.Contains(t, fs.FailedAccountIDs, int64(100))
@@ -229,7 +276,7 @@ func TestHandleFailoverError_CacheBilling(t *testing.T) {
 		fs := NewFailoverState(3, true) // hasBoundSession=true
 		err := newTestFailoverErr(500, false, false)
 
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.True(t, fs.ForceCacheBilling)
 	})
 
@@ -238,7 +285,7 @@ func TestHandleFailoverError_CacheBilling(t *testing.T) {
 		fs := NewFailoverState(3, false)
 		err := newTestFailoverErr(500, false, true) // ForceCacheBilling=true
 
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.True(t, fs.ForceCacheBilling)
 	})
 
@@ -247,7 +294,7 @@ func TestHandleFailoverError_CacheBilling(t *testing.T) {
 		fs := NewFailoverState(3, false)
 		err := newTestFailoverErr(500, false, false)
 
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.False(t, fs.ForceCacheBilling)
 	})
 
@@ -257,12 +304,12 @@ func TestHandleFailoverError_CacheBilling(t *testing.T) {
 
 		// 第一次：ForceCacheBilling=true → 设置
 		err1 := newTestFailoverErr(500, false, true)
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", err1)
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err1)
 		require.True(t, fs.ForceCacheBilling)
 
 		// 第二次：ForceCacheBilling=false → 仍然保持 true
 		err2 := newTestFailoverErr(502, false, false)
-		fs.HandleFailoverError(context.Background(), mock, 200, "openai", err2)
+		fs.HandleFailoverError(context.Background(), mock, 200, "openai", maxSameAccountRetries, err2)
 		require.True(t, fs.ForceCacheBilling, "ForceCacheBilling 一旦设置不应被重置")
 	})
 }
@@ -278,7 +325,7 @@ func TestHandleFailoverError_SameAccountRetry(t *testing.T) {
 		err := newTestFailoverErr(400, true, false)
 
 		start := time.Now()
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		elapsed := time.Since(start)
 
 		require.Equal(t, FailoverContinue, action)
@@ -297,7 +344,7 @@ func TestHandleFailoverError_SameAccountRetry(t *testing.T) {
 		err := newTestFailoverErr(400, true, false)
 
 		for i := 1; i <= maxSameAccountRetries; i++ {
-			action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+			action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 			require.Equal(t, FailoverContinue, action)
 			require.Equal(t, i, fs.SameAccountRetryCount[100])
 		}
@@ -311,12 +358,12 @@ func TestHandleFailoverError_SameAccountRetry(t *testing.T) {
 		err := newTestFailoverErr(400, true, false)
 
 		for i := 0; i < maxSameAccountRetries; i++ {
-			fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+			fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		}
 		require.Equal(t, maxSameAccountRetries, fs.SameAccountRetryCount[100])
 
 		// 第 maxSameAccountRetries+1 次：重试耗尽，应切换账号
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 1, fs.SwitchCount)
 		require.Contains(t, fs.FailedAccountIDs, int64(100))
@@ -333,12 +380,12 @@ func TestHandleFailoverError_SameAccountRetry(t *testing.T) {
 		err := newTestFailoverErr(400, true, false)
 
 		// 账号 100 第一次重试
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 1, fs.SameAccountRetryCount[100])
 
 		// 账号 200 第一次重试（独立计数）
-		action = fs.HandleFailoverError(context.Background(), mock, 200, "openai", err)
+		action = fs.HandleFailoverError(context.Background(), mock, 200, "openai", maxSameAccountRetries, err)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 1, fs.SameAccountRetryCount[200])
 		require.Equal(t, 1, fs.SameAccountRetryCount[100], "账号 100 的计数不应受影响")
@@ -351,17 +398,54 @@ func TestHandleFailoverError_SameAccountRetry(t *testing.T) {
 
 		// 耗尽账号 100 的重试
 		for i := 0; i < maxSameAccountRetries; i++ {
-			fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+			fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		}
 		// 第 maxSameAccountRetries+1 次: 重试耗尽 → 切换
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.Equal(t, FailoverContinue, action)
 
 		// 再次遇到账号 100，计数仍为 maxSameAccountRetries，条件不满足 → 直接切换
-		action = fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		action = fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.Equal(t, FailoverContinue, action)
 		require.Len(t, mock.calls, 2, "第二次耗尽也应调用 TempUnschedule")
 	})
+
+	t.Run("尊重账号级retryLimit_配置1次只重试1次", func(t *testing.T) {
+		// 回归测试：Anthropic 等路径此前硬编码同账号重试 3 次，忽略账号
+		// pool_mode_retry_count 配置。此处验证传入 retryLimit=1 时只重试 1 次即切换。
+		mock := &mockTempUnscheduler{}
+		fs := NewFailoverState(5, false)
+		err := newTestFailoverErr(403, true, false)
+		const retryLimit = 1
+
+		// 第 1 次：同账号重试
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", retryLimit, err)
+		require.Equal(t, FailoverContinue, action)
+		require.Equal(t, 1, fs.SameAccountRetryCount[100])
+		require.Equal(t, 0, fs.SwitchCount, "首次重试不应切换账号")
+		require.Empty(t, mock.calls, "未耗尽前不应 TempUnschedule")
+
+		// 第 2 次：已达上限 1 → 不再同账号重试，直接切换 + TempUnschedule
+		action = fs.HandleFailoverError(context.Background(), mock, 100, "openai", retryLimit, err)
+		require.Equal(t, FailoverContinue, action)
+		require.Equal(t, 1, fs.SameAccountRetryCount[100], "重试计数不应超过 retryLimit")
+		require.Equal(t, 1, fs.SwitchCount, "重试耗尽应切换账号")
+		require.Contains(t, fs.FailedAccountIDs, int64(100))
+		require.Len(t, mock.calls, 1, "重试耗尽应触发 TempUnschedule")
+	})
+
+	t.Run("retryLimit为0时立即切换不重试", func(t *testing.T) {
+		// pool_mode_retry_count=0 表示关闭同账号重试（如 GPT Image 账号）。
+		mock := &mockTempUnscheduler{}
+		fs := NewFailoverState(5, false)
+		err := newTestFailoverErr(403, true, false)
+
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", 0, err)
+		require.Equal(t, FailoverContinue, action)
+		require.Equal(t, 0, fs.SameAccountRetryCount[100], "retryLimit=0 不应发生同账号重试")
+		require.Equal(t, 1, fs.SwitchCount, "应立即切换账号")
+		require.Len(t, mock.calls, 1, "应立即 TempUnschedule")
+	})
 }
 
 // ---------------------------------------------------------------------------
@@ -374,7 +458,7 @@ func TestHandleFailoverError_TempUnschedule(t *testing.T) {
 		fs := NewFailoverState(3, false)
 		err := newTestFailoverErr(500, false, false) // RetryableOnSameAccount=false
 
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.Empty(t, mock.calls)
 	})
 
@@ -384,10 +468,10 @@ func TestHandleFailoverError_TempUnschedule(t *testing.T) {
 		err := newTestFailoverErr(502, true, false)
 
 		for i := 0; i < maxSameAccountRetries; i++ {
-			fs.HandleFailoverError(context.Background(), mock, 42, "openai", err)
+			fs.HandleFailoverError(context.Background(), mock, 42, "openai", maxSameAccountRetries, err)
 		}
 		// 再次触发时才会执行 TempUnschedule + 切换
-		fs.HandleFailoverError(context.Background(), mock, 42, "openai", err)
+		fs.HandleFailoverError(context.Background(), mock, 42, "openai", maxSameAccountRetries, err)
 
 		require.Len(t, mock.calls, 1)
 		require.Equal(t, int64(42), mock.calls[0].accountID)
@@ -407,16 +491,41 @@ func TestHandleFailoverError_ContextCanceled(t *testing.T) {
 		err := newTestFailoverErr(400, true, false)
 
 		ctx, cancel := context.WithCancel(context.Background())
-		cancel() // 立即取消
+		go func() {
+			time.Sleep(30 * time.Millisecond)
+			cancel() // 通过入口检查后、sleep 期间取消
+		}()
 
 		start := time.Now()
-		action := fs.HandleFailoverError(ctx, mock, 100, "openai", err)
+		action := fs.HandleFailoverError(ctx, mock, 100, "openai", maxSameAccountRetries, err)
+		elapsed := time.Since(start)
+
+		require.Equal(t, FailoverCanceled, action)
+		require.Less(t, elapsed, 400*time.Millisecond, "sleep 应被取消打断")
+		// 进入重试分支后才取消：重试计数已递增
+		require.Equal(t, 1, fs.SameAccountRetryCount[100])
+	})
+
+	t.Run("入口即已取消_不改动任何failover状态", func(t *testing.T) {
+		mock := &mockTempUnscheduler{}
+		fs := NewFailoverState(3, false)
+		err := newTestFailoverErr(520, false, false)
+
+		ctx, cancel := context.WithCancel(context.Background())
+		cancel() // 调用前客户端已断开
+
+		start := time.Now()
+		action := fs.HandleFailoverError(ctx, mock, 100, "openai", maxSameAccountRetries, err)
 		elapsed := time.Since(start)
 
 		require.Equal(t, FailoverCanceled, action)
 		require.Less(t, elapsed, 100*time.Millisecond, "应立即返回")
-		// 重试计数仍应递增
-		require.Equal(t, 1, fs.SameAccountRetryCount[100])
+		// 入口已取消时不得改变任何 failover 状态。
+		require.Equal(t, 0, fs.SwitchCount, "取消的请求不应计入切换")
+		require.Equal(t, 0, fs.SameAccountRetryCount[100], "取消的请求不应改动重试计数")
+		require.NotContains(t, fs.FailedAccountIDs, int64(100))
+		require.Nil(t, fs.LastFailoverErr)
+		require.Empty(t, mock.calls, "不应触发 TempUnschedule")
 	})
 
 	t.Run("Antigravity延迟期间context取消", func(t *testing.T) {
@@ -429,7 +538,7 @@ func TestHandleFailoverError_ContextCanceled(t *testing.T) {
 		cancel() // 立即取消
 
 		start := time.Now()
-		action := fs.HandleFailoverError(ctx, mock, 100, service.PlatformAntigravity, err)
+		action := fs.HandleFailoverError(ctx, mock, 100, service.PlatformAntigravity, maxSameAccountRetries, err)
 		elapsed := time.Since(start)
 
 		require.Equal(t, FailoverCanceled, action)
@@ -446,10 +555,10 @@ func TestHandleFailoverError_FailedAccountIDs(t *testing.T) {
 		mock := &mockTempUnscheduler{}
 		fs := NewFailoverState(3, false)
 
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", newTestFailoverErr(500, false, false))
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, newTestFailoverErr(500, false, false))
 		require.Contains(t, fs.FailedAccountIDs, int64(100))
 
-		fs.HandleFailoverError(context.Background(), mock, 200, "openai", newTestFailoverErr(502, false, false))
+		fs.HandleFailoverError(context.Background(), mock, 200, "openai", maxSameAccountRetries, newTestFailoverErr(502, false, false))
 		require.Contains(t, fs.FailedAccountIDs, int64(200))
 		require.Len(t, fs.FailedAccountIDs, 2)
 	})
@@ -458,7 +567,7 @@ func TestHandleFailoverError_FailedAccountIDs(t *testing.T) {
 		mock := &mockTempUnscheduler{}
 		fs := NewFailoverState(0, false)
 
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", newTestFailoverErr(500, false, false))
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, newTestFailoverErr(500, false, false))
 		require.Equal(t, FailoverExhausted, action)
 		require.Contains(t, fs.FailedAccountIDs, int64(100))
 	})
@@ -467,7 +576,7 @@ func TestHandleFailoverError_FailedAccountIDs(t *testing.T) {
 		mock := &mockTempUnscheduler{}
 		fs := NewFailoverState(3, false)
 
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", newTestFailoverErr(400, true, false))
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, newTestFailoverErr(400, true, false))
 		require.Equal(t, FailoverContinue, action)
 		require.NotContains(t, fs.FailedAccountIDs, int64(100))
 	})
@@ -476,8 +585,8 @@ func TestHandleFailoverError_FailedAccountIDs(t *testing.T) {
 		mock := &mockTempUnscheduler{}
 		fs := NewFailoverState(5, false)
 
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", newTestFailoverErr(500, false, false))
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", newTestFailoverErr(500, false, false))
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, newTestFailoverErr(500, false, false))
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, newTestFailoverErr(500, false, false))
 		require.Len(t, fs.FailedAccountIDs, 1, "map 天然去重")
 	})
 }
@@ -492,11 +601,11 @@ func TestHandleFailoverError_LastFailoverErr(t *testing.T) {
 		fs := NewFailoverState(3, false)
 
 		err1 := newTestFailoverErr(500, false, false)
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", err1)
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err1)
 		require.Equal(t, err1, fs.LastFailoverErr)
 
 		err2 := newTestFailoverErr(502, false, false)
-		fs.HandleFailoverError(context.Background(), mock, 200, "openai", err2)
+		fs.HandleFailoverError(context.Background(), mock, 200, "openai", maxSameAccountRetries, err2)
 		require.Equal(t, err2, fs.LastFailoverErr)
 	})
 
@@ -505,7 +614,7 @@ func TestHandleFailoverError_LastFailoverErr(t *testing.T) {
 		fs := NewFailoverState(3, false)
 
 		err := newTestFailoverErr(400, true, false)
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.Equal(t, err, fs.LastFailoverErr)
 	})
 }
@@ -522,30 +631,30 @@ func TestHandleFailoverError_IntegrationScenario(t *testing.T) {
 		// 1. 账号 100 遇到可重试错误，同账号重试 maxSameAccountRetries 次
 		retryErr := newTestFailoverErr(400, true, false)
 		for i := 0; i < maxSameAccountRetries; i++ {
-			action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", retryErr)
+			action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, retryErr)
 			require.Equal(t, FailoverContinue, action)
 		}
 		require.True(t, fs.ForceCacheBilling, "hasBoundSession=true 应设置 ForceCacheBilling")
 
 		// 2. 账号 100 超过重试上限 → TempUnschedule + 切换
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", retryErr)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, retryErr)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 1, fs.SwitchCount)
 		require.Len(t, mock.calls, 1)
 
 		// 3. 账号 200 遇到不可重试错误 → 直接切换
 		switchErr := newTestFailoverErr(500, false, false)
-		action = fs.HandleFailoverError(context.Background(), mock, 200, "openai", switchErr)
+		action = fs.HandleFailoverError(context.Background(), mock, 200, "openai", maxSameAccountRetries, switchErr)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 2, fs.SwitchCount)
 
 		// 4. 账号 300 遇到不可重试错误 → 再切换
-		action = fs.HandleFailoverError(context.Background(), mock, 300, "openai", switchErr)
+		action = fs.HandleFailoverError(context.Background(), mock, 300, "openai", maxSameAccountRetries, switchErr)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 3, fs.SwitchCount)
 
 		// 5. 账号 400 → 已耗尽 (SwitchCount=3 >= MaxSwitches=3)
-		action = fs.HandleFailoverError(context.Background(), mock, 400, "openai", switchErr)
+		action = fs.HandleFailoverError(context.Background(), mock, 400, "openai", maxSameAccountRetries, switchErr)
 		require.Equal(t, FailoverExhausted, action)
 
 		// 最终状态验证
@@ -563,21 +672,21 @@ func TestHandleFailoverError_IntegrationScenario(t *testing.T) {
 
 		// 第一次切换：delay = 0s
 		start := time.Now()
-		action := fs.HandleFailoverError(context.Background(), mock, 100, service.PlatformAntigravity, err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, service.PlatformAntigravity, maxSameAccountRetries, err)
 		elapsed := time.Since(start)
 		require.Equal(t, FailoverContinue, action)
 		require.Less(t, elapsed, 200*time.Millisecond, "第一次切换延迟为 0")
 
 		// 第二次切换：delay = 1s
 		start = time.Now()
-		action = fs.HandleFailoverError(context.Background(), mock, 200, service.PlatformAntigravity, err)
+		action = fs.HandleFailoverError(context.Background(), mock, 200, service.PlatformAntigravity, maxSameAccountRetries, err)
 		elapsed = time.Since(start)
 		require.Equal(t, FailoverContinue, action)
 		require.GreaterOrEqual(t, elapsed, 800*time.Millisecond, "第二次切换延迟约 1s")
 
 		// 第三次：耗尽（无延迟，因为在检查延迟之前就返回了）
 		start = time.Now()
-		action = fs.HandleFailoverError(context.Background(), mock, 300, service.PlatformAntigravity, err)
+		action = fs.HandleFailoverError(context.Background(), mock, 300, service.PlatformAntigravity, maxSameAccountRetries, err)
 		elapsed = time.Since(start)
 		require.Equal(t, FailoverExhausted, action)
 		require.Less(t, elapsed, 200*time.Millisecond, "耗尽时不应有延迟")
@@ -589,17 +698,17 @@ func TestHandleFailoverError_IntegrationScenario(t *testing.T) {
 
 		// 第一次：ForceCacheBilling=false
 		err1 := newTestFailoverErr(500, false, false)
-		fs.HandleFailoverError(context.Background(), mock, 100, "openai", err1)
+		fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err1)
 		require.False(t, fs.ForceCacheBilling)
 
 		// 第二次：ForceCacheBilling=true（Antigravity 粘性会话切换）
 		err2 := newTestFailoverErr(500, false, true)
-		fs.HandleFailoverError(context.Background(), mock, 200, "openai", err2)
+		fs.HandleFailoverError(context.Background(), mock, 200, "openai", maxSameAccountRetries, err2)
 		require.True(t, fs.ForceCacheBilling, "错误标志应触发 ForceCacheBilling")
 
 		// 第三次：ForceCacheBilling=false，但状态仍保持 true
 		err3 := newTestFailoverErr(500, false, false)
-		fs.HandleFailoverError(context.Background(), mock, 300, "openai", err3)
+		fs.HandleFailoverError(context.Background(), mock, 300, "openai", maxSameAccountRetries, err3)
 		require.True(t, fs.ForceCacheBilling, "不应重置")
 	})
 }
@@ -614,7 +723,7 @@ func TestHandleFailoverError_EdgeCases(t *testing.T) {
 		fs := NewFailoverState(3, false)
 		err := newTestFailoverErr(0, false, false)
 
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "openai", maxSameAccountRetries, err)
 		require.Equal(t, FailoverContinue, action)
 	})
 
@@ -623,7 +732,7 @@ func TestHandleFailoverError_EdgeCases(t *testing.T) {
 		fs := NewFailoverState(3, false)
 		err := newTestFailoverErr(500, true, false)
 
-		action := fs.HandleFailoverError(context.Background(), mock, 0, "openai", err)
+		action := fs.HandleFailoverError(context.Background(), mock, 0, "openai", maxSameAccountRetries, err)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 1, fs.SameAccountRetryCount[0])
 	})
@@ -633,7 +742,7 @@ func TestHandleFailoverError_EdgeCases(t *testing.T) {
 		fs := NewFailoverState(3, false)
 		err := newTestFailoverErr(500, true, false)
 
-		action := fs.HandleFailoverError(context.Background(), mock, -1, "openai", err)
+		action := fs.HandleFailoverError(context.Background(), mock, -1, "openai", maxSameAccountRetries, err)
 		require.Equal(t, FailoverContinue, action)
 		require.Equal(t, 1, fs.SameAccountRetryCount[-1])
 	})
@@ -645,7 +754,7 @@ func TestHandleFailoverError_EdgeCases(t *testing.T) {
 		err := newTestFailoverErr(500, false, false)
 
 		start := time.Now()
-		action := fs.HandleFailoverError(context.Background(), mock, 100, "", err)
+		action := fs.HandleFailoverError(context.Background(), mock, 100, "", maxSameAccountRetries, err)
 		elapsed := time.Since(start)
 
 		require.Equal(t, FailoverContinue, action)
@@ -718,6 +827,29 @@ func TestHandleSelectionExhausted(t *testing.T) {
 		require.Less(t, elapsed, 100*time.Millisecond, "应立即返回")
 	})
 
+	t.Run("context已取消_非503也返回Canceled而非Exhausted", func(t *testing.T) {
+		// #4257 核心场景：客户端断开后选号失败源于 context canceled，
+		// 不应被当成账号耗尽转成 502。
+		fs := NewFailoverState(3, false)
+		fs.LastFailoverErr = newTestFailoverErr(520, false, false)
+
+		ctx, cancel := context.WithCancel(context.Background())
+		cancel()
+
+		action := fs.HandleSelectionExhausted(ctx)
+		require.Equal(t, FailoverCanceled, action)
+	})
+
+	t.Run("context已取消_无LastFailoverErr也返回Canceled", func(t *testing.T) {
+		fs := NewFailoverState(3, false)
+
+		ctx, cancel := context.WithCancel(context.Background())
+		cancel()
+
+		action := fs.HandleSelectionExhausted(ctx)
+		require.Equal(t, FailoverCanceled, action)
+	})
+
 	t.Run("503且SwitchCount等于MaxSwitches_仍可重试", func(t *testing.T) {
 		fs := NewFailoverState(2, false)
 		fs.LastFailoverErr = newTestFailoverErr(503, false, false)
@@ -727,3 +859,47 @@ func TestHandleSelectionExhausted(t *testing.T) {
 		require.Equal(t, FailoverContinue, action)
 	})
 }
+
+// ---------------------------------------------------------------------------
+// failoverClientGone 测试
+// ---------------------------------------------------------------------------
+
+func TestFailoverClientGone(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+
+	t.Run("活跃请求返回false", func(t *testing.T) {
+		rec := httptest.NewRecorder()
+		c, _ := gin.CreateTestContext(rec)
+		c.Request = httptest.NewRequest(http.MethodPost, "/v1/responses", nil)
+
+		require.False(t, failoverClientGone(c))
+		require.Equal(t, http.StatusOK, c.Writer.Status(), "不应改动状态码")
+	})
+
+	t.Run("客户端已断开_返回true并标记499", func(t *testing.T) {
+		rec := httptest.NewRecorder()
+		c, _ := gin.CreateTestContext(rec)
+		ctx, cancel := context.WithCancel(context.Background())
+		cancel()
+		c.Request = httptest.NewRequest(http.MethodPost, "/v1/responses", nil).WithContext(ctx)
+
+		require.True(t, failoverClientGone(c))
+		require.Equal(t, statusClientClosedRequest, c.Writer.Status())
+	})
+
+	t.Run("响应已提交_不改状态码", func(t *testing.T) {
+		rec := httptest.NewRecorder()
+		c, _ := gin.CreateTestContext(rec)
+		ctx, cancel := context.WithCancel(context.Background())
+		cancel()
+		c.Request = httptest.NewRequest(http.MethodPost, "/v1/responses", nil).WithContext(ctx)
+		c.String(http.StatusOK, "partial")
+
+		require.True(t, failoverClientGone(c))
+		require.Equal(t, http.StatusOK, c.Writer.Status(), "已提交的状态码不应被覆盖")
+	})
+
+	t.Run("nil安全", func(t *testing.T) {
+		require.False(t, failoverClientGone(nil))
+	})
+}

backend/internal/handler/gateway_handler.go

@@ -39,6 +39,7 @@ var gatewayCompatibilityMetricsLogCounter atomic.Uint64
 // GatewayHandler handles API gateway requests
 type GatewayHandler struct {
 	gatewayService            *service.GatewayService
+	openAIGatewayService      *service.OpenAIGatewayService
 	geminiCompatService       *service.GeminiMessagesCompatService
 	antigravityGatewayService *service.AntigravityGatewayService
 	userService               *service.UserService
@@ -59,6 +60,7 @@ type GatewayHandler struct {
 // NewGatewayHandler creates a new GatewayHandler
 func NewGatewayHandler(
 	gatewayService *service.GatewayService,
+	openAIGatewayService *service.OpenAIGatewayService,
 	geminiCompatService *service.GeminiMessagesCompatService,
 	antigravityGatewayService *service.AntigravityGatewayService,
 	userService *service.UserService,
@@ -94,6 +96,7 @@ func NewGatewayHandler(
 
 	return &GatewayHandler{
 		gatewayService:            gatewayService,
+		openAIGatewayService:      openAIGatewayService,
 		geminiCompatService:       geminiCompatService,
 		antigravityGatewayService: antigravityGatewayService,
 		userService:               userService,
@@ -327,6 +330,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 					c.Request = c.Request.WithContext(ctx)
 					continue
 				case FailoverCanceled:
+					failoverClientGone(c)
 					return
 				default: // FailoverExhausted
 					if fs.LastFailoverErr != nil {
@@ -448,7 +452,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 						h.handleFailoverExhausted(c, failoverErr, service.PlatformGemini, true)
 						return
 					}
-					action := fs.HandleFailoverError(c.Request.Context(), h.gatewayService, account.ID, account.Platform, failoverErr)
+					action := fs.HandleFailoverError(c.Request.Context(), h.gatewayService, account.ID, account.Platform, account.GetPoolModeRetryCount(), failoverErr)
 					switch action {
 					case FailoverContinue:
 						continue
@@ -456,6 +460,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 						h.handleFailoverExhausted(c, fs.LastFailoverErr, service.PlatformGemini, streamStarted)
 						return
 					case FailoverCanceled:
+						failoverClientGone(c)
 						return
 					}
 				}
@@ -613,6 +618,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 					c.Request = c.Request.WithContext(ctx)
 					continue
 				case FailoverCanceled:
+					failoverClientGone(c)
 					return
 				default: // FailoverExhausted
 					if fs.LastFailoverErr != nil {
@@ -868,7 +874,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 						h.handleFailoverExhausted(c, failoverErr, account.Platform, true)
 						return
 					}
-					action := fs.HandleFailoverError(c.Request.Context(), h.gatewayService, account.ID, account.Platform, failoverErr)
+					action := fs.HandleFailoverError(c.Request.Context(), h.gatewayService, account.ID, account.Platform, account.GetPoolModeRetryCount(), failoverErr)
 					switch action {
 					case FailoverContinue:
 						continue
@@ -876,6 +882,7 @@ func (h *GatewayHandler) Messages(c *gin.Context) {
 						h.handleFailoverExhausted(c, fs.LastFailoverErr, account.Platform, streamStarted)
 						return
 					case FailoverCanceled:
+						failoverClientGone(c)
 						return
 					}
 				}
@@ -1454,13 +1461,14 @@ func (h *GatewayHandler) usageUnrestricted(c *gin.Context, ctx context.Context,
 			remaining := h.calculateSubscriptionRemaining(apiKey.Group, subscription)
 			resp["remaining"] = remaining
 			resp["subscription"] = gin.H{
-				"daily_usage_usd":   subscription.DailyUsageUSD,
-				"weekly_usage_usd":  subscription.WeeklyUsageUSD,
-				"monthly_usage_usd": subscription.MonthlyUsageUSD,
-				"daily_limit_usd":   apiKey.Group.DailyLimitUSD,
-				"weekly_limit_usd":  apiKey.Group.WeeklyLimitUSD,
-				"monthly_limit_usd": apiKey.Group.MonthlyLimitUSD,
-				"expires_at":        subscription.ExpiresAt,
+				"daily_usage_usd":     subscription.DailyUsageUSD,
+				"weekly_usage_usd":    subscription.WeeklyUsageUSD,
+				"monthly_usage_usd":   subscription.MonthlyUsageUSD,
+				"daily_limit_usd":     apiKey.Group.DailyLimitUSD,
+				"weekly_limit_usd":    apiKey.Group.WeeklyLimitUSD,
+				"monthly_limit_usd":   apiKey.Group.MonthlyLimitUSD,
+				"weekly_window_start": subscription.WeeklyWindowStart,
+				"expires_at":          subscription.ExpiresAt,
 			}
 		}
 

backend/internal/handler/gateway_handler_cancellation_test.go

@@ -0,0 +1,98 @@
+//go:build unit
+
+package handler
+
+import (
+	"bytes"
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"sync/atomic"
+	"testing"
+
+	"github.com/Wei-Shaw/sub2api/internal/config"
+	"github.com/Wei-Shaw/sub2api/internal/pkg/ctxkey"
+	middleware "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"
+)
+
+type countingGatewaySchedulerCache struct {
+	*fakeSchedulerCache
+	snapshotCalls atomic.Int64
+}
+
+func (c *countingGatewaySchedulerCache) GetSnapshot(ctx context.Context, bucket service.SchedulerBucket) ([]*service.Account, bool, error) {
+	c.snapshotCalls.Add(1)
+	return c.fakeSchedulerCache.GetSnapshot(ctx, bucket)
+}
+
+func TestGatewayHandlerPreCancelledCompatibleRequestsDoNotSelectAccount(t *testing.T) {
+	gin.SetMode(gin.TestMode)
+	groupID := int64(9100)
+	group := &service.Group{ID: groupID, Hydrated: true, Platform: service.PlatformAnthropic, Status: service.StatusActive}
+	account := &service.Account{
+		ID: 9101, Platform: service.PlatformAnthropic, Type: service.AccountTypeAPIKey,
+		Status: service.StatusActive, Schedulable: true, Concurrency: 1,
+		AccountGroups: []service.AccountGroup{{AccountID: 9101, GroupID: groupID}},
+	}
+	schedulerCache := &countingGatewaySchedulerCache{fakeSchedulerCache: &fakeSchedulerCache{accounts: []*service.Account{account}}}
+	schedulerSnapshot := service.NewSchedulerSnapshotService(schedulerCache, nil, nil, nil, nil)
+	gatewayService := service.NewGatewayService(
+		nil, &fakeGroupRepo{group: group}, nil, nil, nil, nil, nil, nil, nil,
+		schedulerSnapshot, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil,
+	)
+	cfg := &config.Config{RunMode: config.RunModeSimple}
+	billingCacheService := service.NewBillingCacheService(nil, nil, nil, nil, nil, nil, cfg, nil)
+	t.Cleanup(billingCacheService.Stop)
+	h := &GatewayHandler{
+		gatewayService:      gatewayService,
+		billingCacheService: billingCacheService,
+		concurrencyHelper:   NewConcurrencyHelper(service.NewConcurrencyService(&fakeConcurrencyCache{}), SSEPingFormatClaude, 0),
+		maxAccountSwitches:  1,
+		cfg:                 cfg,
+	}
+	apiKey := &service.APIKey{
+		ID: 9102, UserID: 9103, GroupID: &groupID, Group: group, Status: service.StatusActive,
+		User: &service.User{ID: 9103, Concurrency: 10, Balance: 100},
+	}
+
+	tests := []struct {
+		name string
+		path string
+		body string
+		call func(*gin.Context)
+	}{
+		{
+			name: "responses", path: "/v1/responses", body: `{"model":"claude-test","input":"hello","stream":false}`,
+			call: h.Responses,
+		},
+		{
+			name: "chat completions", path: "/v1/chat/completions", body: `{"model":"claude-test","messages":[{"role":"user","content":"hello"}],"stream":false}`,
+			call: h.ChatCompletions,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			schedulerCache.snapshotCalls.Store(0)
+			recorder := httptest.NewRecorder()
+			c, _ := gin.CreateTestContext(recorder)
+			ctx, cancel := context.WithCancel(context.Background())
+			cancel()
+			ctx = context.WithValue(ctx, ctxkey.Group, group)
+			req := httptest.NewRequest(http.MethodPost, tt.path, bytes.NewBufferString(tt.body)).WithContext(ctx)
+			req.Header.Set("Content-Type", "application/json")
+			c.Request = req
+			c.Set(string(middleware.ContextKeyAPIKey), apiKey)
+			c.Set(string(middleware.ContextKeyUser), middleware.AuthSubject{UserID: apiKey.UserID, Concurrency: 10})
+
+			tt.call(c)
+
+			require.Zero(t, schedulerCache.snapshotCalls.Load(), "a cancelled request must stop before the account selector")
+			_, selected := c.Get(opsAccountIDKey)
+			require.False(t, selected)
+		})
+	}
+}

backend/internal/handler/gateway_handler_chat_completions.go

@@ -80,6 +80,10 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 		h.chatCompletionsErrorResponse(c, http.StatusBadRequest, "invalid_request_error", invalidStreamFieldTypeMessage)
 		return
 	}
+	if service.IsGPTImageGenerationModel(reqModel) {
+		h.chatCompletionsErrorResponse(c, http.StatusBadRequest, "invalid_request_error", "This model is not supported on the Chat Completions endpoint")
+		return
+	}
 	reqLog = reqLog.With(zap.String("model", reqModel), zap.Bool("stream", reqStream))
 
 	setOpsRequestContext(c, reqModel, reqStream)
@@ -159,6 +163,9 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 	}
 
 	for {
+		if c.Request.Context().Err() != nil {
+			return
+		}
 		selection, err := h.gatewayService.SelectAccountWithLoadAwareness(c.Request.Context(), apiKey.GroupID, selectionSessionHash, reqModel, fs.FailedAccountIDs, "", int64(0))
 		if err != nil {
 			if len(fs.FailedAccountIDs) == 0 {
@@ -178,6 +185,7 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 			case FailoverContinue:
 				continue
 			case FailoverCanceled:
+				failoverClientGone(c)
 				return
 			default:
 				if fs.LastFailoverErr != nil {
@@ -254,7 +262,7 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 					h.handleCCFailoverExhausted(c, failoverErr, true)
 					return
 				}
-				action := fs.HandleFailoverError(c.Request.Context(), h.gatewayService, account.ID, account.Platform, failoverErr)
+				action := fs.HandleFailoverError(c.Request.Context(), h.gatewayService, account.ID, account.Platform, account.GetPoolModeRetryCount(), failoverErr)
 				switch action {
 				case FailoverContinue:
 					continue
@@ -262,6 +270,7 @@ func (h *GatewayHandler) ChatCompletions(c *gin.Context) {
 					h.handleCCFailoverExhausted(c, fs.LastFailoverErr, streamStarted)
 					return
 				case FailoverCanceled:
+					failoverClientGone(c)
 					return
 				}
 			}
@@ -328,6 +337,14 @@ func (h *GatewayHandler) handleCCFailoverExhausted(c *gin.Context, lastErr *serv
 	if streamStarted {
 		return
 	}
+	if lastErr != nil {
+		copyFailoverRetryAfter(c, lastErr.ResponseHeaders)
+	}
+	if lastErr != nil && lastErr.IsCredentialFailure() {
+		status, message := credentialFailoverClientResponse(lastErr)
+		h.chatCompletionsErrorResponse(c, status, "server_error", message)
+		return
+	}
 	statusCode := http.StatusBadGateway
 	if lastErr != nil && lastErr.StatusCode > 0 {
 		statusCode = lastErr.StatusCode