feeeei 26b31a83e4 fix(gemini): 显式缓存修改串行化、失败回滚，确定金额费用按 Key / 订阅预留 ...

同一缓存的并发 PATCH 可能以与上游提交不同的顺序写回本地到期时间；
确定金额的存储费只预留用户余额，并发续期可越过 API Key 总额度与订阅额度；
上游已修改而本地写回失败时没有可恢复的状态。

- 新增迁移 245：缓存记录增加修改意图（mutation_token / mutation_lease_until）。
  PATCH 在调用上游前登记意图并取得 2 分钟租约，条件为没有进行中的修改且到期时间
  未变；上游修改、本地写回与计费均由持有者完成，并发的修改返回 409 ABORTED
- 上游结果未知、本地写回失败或进程退出时修改意图保留且不计费；租约过期后由下一次
  修改或后台每分钟的恢复流程把上游有效期回滚到修改前的到期时间（已到期则删除上游
  缓存与本地记录）。上游明确拒绝或请求未发出时直接清除意图
- 修改 / 删除上游缓存的单次调用超时改为 30 秒；请求发出前的失败包装为
  ErrGeminiCachedContentRequestNotSent
- CheckChargeAffordable 与 ReserveInflightCharge 合并为 ReserveFixedCharge：在途预留
  新增订阅、API Key 两个维度（Redis billing:inflight_scope:*），先登记本笔预留再读取
  库中权威用量（Key quota_used 与限速窗口、订阅日 / 周 / 月用量）叠加同维度在途合计
  校验；预留随计费任务交接，落库后释放
- GetRateLimitData 一并读取 quota_used

2026-10-12 17:14:56 +08:00

..

admin

fix(provider): 修复 Claude 计费与钱包冷却并隔离模型目录缓存

2026-10-09 10:11:06 +08:00

dto

feat(payment): 充值优惠阶梯支持折扣模式，快捷金额角标改为促销价签

2026-10-02 00:25:41 +08:00

quotaview

…

announcement_handler.go

…

api_key_handler_validation_test.go

fix(api-key): validate quota and expiry inputs

2026-08-09 22:42:51 +08:00

api_key_handler.go

fix(api-key): validate quota and expiry inputs

2026-08-09 22:42:51 +08:00

auth_captcha_request_test.go

fix: 完善腾讯验证码区域适配与 CSP 白名单

2026-08-06 20:34:37 +08:00

auth_current_user_test.go

feat(groups)!: enforce model allowlists across gateway endpoints

2026-09-05 16:05:17 +08:00

auth_dingtalk_client_test.go

…

auth_dingtalk_client.go

…

auth_dingtalk_oauth_test.go

…

auth_dingtalk_oauth.go

人机验证增加阿里云验证码 2.0

2026-08-04 20:57:15 +08:00

auth_email_oauth_test.go

fix(affiliate): 线下提现登记按 Idempotency-Key 幂等

2026-09-23 01:24:59 +08:00

auth_email_oauth.go

人机验证增加阿里云验证码 2.0

2026-08-04 20:57:15 +08:00

auth_handler.go

新增腾讯天御验证码认证门禁

2026-08-04 15:09:29 +08:00

auth_linuxdo_oauth_test.go

…

auth_linuxdo_oauth.go

人机验证增加阿里云验证码 2.0

2026-08-04 20:57:15 +08:00

auth_oauth_captcha_start_test.go

人机验证增加阿里云验证码 2.0

2026-08-04 20:57:15 +08:00

auth_oauth_captcha_start.go

人机验证增加阿里云验证码 2.0

2026-08-04 20:57:15 +08:00

auth_oauth_logout_test.go

…

auth_oauth_pending_flow_test.go

fix(email): atomic verify-code attempts and hashed single-use reset tokens

2026-09-28 12:26:28 -04:00

auth_oauth_pending_flow.go

fix(security): block OAuth account takeover via pending exchange

2026-08-07 03:09:57 +08:00

auth_oauth_test_helpers_test.go

…

auth_oidc_oauth_test.go

…

auth_oidc_oauth.go

build: 升级 Go 1.27.0，同步 CI/Dockerfile 并适配 jsonv2 与 golangci-lint v2.13

2026-08-24 12:02:53 +08:00

auth_session_revocation_test.go

refactor(repository): scope user/api-key updates to declared columns

2026-07-28 17:21:32 +08:00

auth_wechat_oauth_test.go

…

auth_wechat_oauth.go

人机验证增加阿里云验证码 2.0

2026-08-04 20:57:15 +08:00

available_channel_handler_test.go

feat(billing): support configurable reasoning effort multipliers

2026-09-20 16:41:49 +08:00

available_channel_handler.go

feat(billing): support configurable reasoning effort multipliers

2026-09-20 16:41:49 +08:00

batch_image_handler.go

feat(groups)!: enforce model allowlists across gateway endpoints

2026-09-05 16:05:17 +08:00

channel_monitor_user_handler.go

feat(monitor): wire quota fetcher & expose check_mode in handlers

2026-08-18 04:14:15 +00:00

channel_monitor_v2_handler_test.go

fix(monitor): scope groups by user access

2026-08-30 09:28:40 +08:00

channel_monitor_v2_handler.go

fix(monitor): scope groups by user access

2026-08-30 09:28:40 +08:00

chat_completions_image_model_test.go

…

composite_platform_test.go

Merge pull request #6758 from DuskLin/feat/minimax-platform

2026-09-08 18:50:06 +08:00

composite_platform.go

feat: add Anthropic reasoning effort pricing controls

2026-09-02 20:16:57 +08:00

concurrency_error_response_test.go

fix(gateway): 客户端断开统一标记 499,不再被记成 502/200 上游错误

2026-09-26 17:13:10 +08:00

concurrency_error_response.go

[verified] fix(handler): distinguish gateway admission rejections

2026-09-03 12:48:02 +08:00

content_moderation_helper.go

…

endpoint_test.go

feat(gemini): 支持显式上下文缓存（AI Studio / Vertex / 自建网关上游）

2026-10-10 23:37:47 +08:00

endpoint.go

feat(gemini): 支持显式上下文缓存（AI Studio / Vertex / 自建网关上游）

2026-10-10 23:37:47 +08:00

failover_loop_profit_veto_test.go

fix(profit-control): stop leaking profit policy, close veto livelock, restore passthrough turn pricing

2026-08-01 22:39:33 +08:00

failover_loop_test.go

修复 5888 与 5925 的同号重试语义冲突

2026-08-20 18:16:38 -07:00

failover_loop.go

修复 5888 与 5925 的同号重试语义冲突

2026-08-20 18:16:38 -07:00

gateway_handler_billing_error_test.go

…

gateway_handler_cancellation_test.go

fix(gateway): 客户端断开统一标记 499,不再被记成 502/200 上游错误

2026-09-26 17:13:10 +08:00

gateway_handler_chat_completions.go

fix(typesafe): 补齐平台迁移、审计覆盖、端点隔离与错误策略

2026-10-02 13:08:40 +08:00

gateway_handler_claude_code_fallback_test.go

fix(gateway): Claude Code 限制分组的 OpenAI 兼容入口改走降级分组,不再直接 403

2026-09-24 16:20:28 +08:00

gateway_handler_error_fallback_test.go

fix(gateway): 客户端断开统一标记 499,不再被记成 502/200 上游错误

2026-09-26 17:13:10 +08:00

gateway_handler_intercept_test.go

fix(gateway): 修复模拟响应使用 Bedrock msg_bdrk_ 格式，改为正宗 Anthropic msg_01 格式

2026-07-27 15:20:03 +00:00

gateway_handler_responses.go

fix(typesafe): 补齐平台迁移、审计覆盖、端点隔离与错误策略

2026-10-02 13:08:40 +08:00

gateway_handler_stream_failover_test.go

…

gateway_handler_usage_test.go

…

gateway_handler_warmup_intercept_unit_test.go

fix(gateway): 修复模拟响应使用 Bedrock msg_bdrk_ 格式，改为正宗 Anthropic msg_01 格式

2026-07-27 15:20:03 +00:00

gateway_handler.go

feat(gemini): 支持显式上下文缓存（AI Studio / Vertex / 自建网关上游）

2026-10-10 23:37:47 +08:00

gateway_helper_backoff_test.go

…

gateway_helper_fastpath_test.go

…

gateway_helper_hotpath_test.go

fix(gateway): accept Claude Code max_tokens=1 probes for any model

2026-09-08 19:12:37 +08:00

gateway_helper_test.go

…

gateway_helper.go

fix(gateway): accept Claude Code max_tokens=1 probes for any model

2026-09-08 19:12:37 +08:00

gateway_inflight_reservation_test.go

fix(billing): hold inflight reservation until billing lands; price estimate like billing; wire remaining endpoints

2026-09-28 16:14:43 -04:00

gateway_inflight_reservation.go

fix(billing): hold inflight reservation until billing lands; price estimate like billing; wire remaining endpoints

2026-09-28 16:14:43 -04:00

gateway_key_billing_test.go

…

gateway_key_billing.go

…

gateway_model_allowlist_listing_test.go

fix(gemini): discover antigravity mappings in native model lists

2026-09-15 16:50:46 +08:00

gateway_models_retrieve_test.go

feat: add native TypeSafe Jev System One support

2026-09-20 20:20:16 +08:00

gateway_models_test.go

fix(models): preserve composite catalog isolation after base update

2026-10-03 09:11:53 +08:00

gateway_systemone_test.go

fix(typesafe): 补齐平台迁移、审计覆盖、端点隔离与错误策略

2026-10-02 13:08:40 +08:00

gateway_systemone.go

fix(typesafe): 堵住校验解析差异与审计盲区，收紧模型列表口径

2026-10-02 13:51:58 +08:00

gateway_web_search.go

fix(billing): hold inflight reservation until billing lands; price estimate like billing; wire remaining endpoints

2026-09-28 16:14:43 -04:00

gemini_cached_contents_billing_test.go

fix(gemini): 显式缓存修改串行化、失败回滚，确定金额费用按 Key / 订阅预留

2026-10-12 17:14:56 +08:00

gemini_cached_contents_handler_test.go

fix(gemini): 显式缓存修改串行化、失败回滚，确定金额费用按 Key / 订阅预留

2026-10-12 17:14:56 +08:00

gemini_cached_contents_handler.go

fix(gemini): 显式缓存修改串行化、失败回滚，确定金额费用按 Key / 订阅预留

2026-10-12 17:14:56 +08:00

gemini_cli_session_test.go

…

gemini_client_cancel_test.go

fix(gateway): 客户端断开统一标记 499,不再被记成 502/200 上游错误

2026-09-26 17:13:10 +08:00

gemini_mixed_models_test.go

fix(gemini): discover antigravity mappings in native model lists

2026-09-15 16:50:46 +08:00

gemini_v1beta_handler_test.go

fix(gemini): 模型列表与模型详情接口支持 Agent Platform (Vertex) 账号

2026-10-11 22:03:36 +08:00

gemini_v1beta_handler.go

fix(gemini): 模型列表与模型详情接口支持 Agent Platform (Vertex) 账号

2026-10-11 22:03:36 +08:00

gemini_vertex_models_test.go

fix(gemini): 模型列表与模型详情接口支持 Agent Platform (Vertex) 账号

2026-10-11 22:03:36 +08:00

grok_audio_billing_test.go

fix: Realtime 仅在观察到音频后计费，并修正标志位求值顺序

2026-08-13 08:38:10 +08:00

grok_audio.go

fix(billing): hold inflight reservation until billing lands; price estimate like billing; wire remaining endpoints

2026-09-28 16:14:43 -04:00

grok_media_slots_test.go

feat: support native Seedance Ark video task API

2026-09-17 04:14:14 +09:00

grok_media_test.go

fix(grok): 视频在 status 返回 video.url 后再按上游参数计费

2026-08-07 16:29:57 +08:00

grok_media.go

fix(billing): hold inflight reservation until billing lands; price estimate like billing; wire remaining endpoints

2026-09-28 16:14:43 -04:00

handler.go

feat: add OAuth outbound transport plugin system

2026-08-24 09:03:37 +08:00

idempotency_helper_test.go

…

idempotency_helper.go

…

image_concurrency_limiter_test.go

…

image_concurrency_limiter.go

…

image_task_admin_toggle_test.go

…

image_task_handler_test.go

…

image_task_handler.go

…

logging.go

…

model_plaza_handler_test.go

fix(model-plaza): apply independent video billing multipliers

2026-09-23 11:28:41 +08:00

model_plaza_handler.go

fix(model-plaza): apply independent video billing multipliers

2026-09-23 11:28:41 +08:00

no_account_error_test.go

fix(gateway): stop a per-model cooldown from turning 404 model_not_found into 429

2026-09-01 15:18:30 +08:00

no_account_error.go

fix(gateway): stop a per-model cooldown from turning 404 model_not_found into 429

2026-09-01 15:18:30 +08:00

openai_alpha_search.go

fix(billing): hold inflight reservation until billing lands; price estimate like billing; wire remaining endpoints

2026-09-28 16:14:43 -04:00

openai_automation_bootstrap_test.go

fix(openai): accept full heartbeat bootstrap envelopes

2026-09-09 12:48:57 +08:00

openai_body_limit_failover_test.go

…

openai_chat_completions.go

fix(billing): hold inflight reservation until billing lands; price estimate like billing; wire remaining endpoints

2026-09-28 16:14:43 -04:00

openai_client_disconnect_test.go

fix(openai-ws): classify canceled client streams

2026-08-31 21:04:29 +08:00

openai_codex_models_handler_test.go

fix(models): enforce account mappings for remote Codex discovery

2026-10-01 03:30:56 -07:00

openai_codex_models_handler.go

feat(openai): reuse pinned accounts for standard model discovery

2026-09-05 19:58:47 +08:00

openai_cyber_allowlist_test.go

feat: add risk control user allowlist

2026-09-29 01:50:21 +08:00

openai_cyber_allowlist.go

feat: add risk control user allowlist

2026-09-29 01:50:21 +08:00

openai_delegation_bootstrap_test.go

fix(openai): 兼容历史任务委派续聊

2026-09-03 11:39:23 +08:00

openai_embeddings.go

fix(billing): hold inflight reservation until billing lands; price estimate like billing; wire remaining endpoints

2026-09-28 16:14:43 -04:00

openai_gateway_cn_dispatch_test.go

fix(opencode): address review on dispatch, billing, session, and 402

2026-09-11 15:39:13 +08:00

openai_gateway_compact_body_signal_test.go

fix(openai): separate native and legacy compaction routing

2026-08-14 18:18:53 +08:00

openai_gateway_compact_log_test.go

fix(openai): separate native and legacy compaction routing

2026-08-14 18:18:53 +08:00

openai_gateway_count_tokens.go

修复 PR 5888 剩余审计问题

2026-08-20 22:01:20 -07:00

openai_gateway_credential_failover_loop_test.go

…

openai_gateway_credential_failover_test.go

fix(gateway): fail over model-not-found responses

2026-09-06 20:38:17 +08:00

openai_gateway_cyber_test.go

fix(openai): record WS cyber policy failures across paths

2026-09-05 15:32:29 +08:00

openai_gateway_endpoint_normalization_test.go

…

openai_gateway_first_output_timeout_test.go

…

openai_gateway_handler_test.go

fix(openai-ws): bill each later turn with the current group pricing

2026-10-01 11:27:35 +08:00

openai_gateway_handler.go

refactor(platform): 后端平台列表改由平台清单派生

2026-10-09 10:10:33 +08:00

openai_gateway_reasoning_failover_test.go

…

openai_gateway_reasoning_failover.go

…

openai_gateway_usage_context_test.go

…

openai_gateway_ws_composite_test.go

fix(gateway): resolve composite routes for websocket aliases

2026-09-23 15:29:11 +08:00

openai_gateway_ws_failover_resume_test.go

fix(openai): resume later websocket turns after 429

2026-08-19 18:51:21 +08:00

openai_gateway_ws_group_pricing_test.go

fix(openai-ws): bill each later turn with the current group pricing

2026-10-01 11:27:35 +08:00

openai_gateway_ws_model_allowlist_test.go

feat(groups)!: enforce model allowlists across gateway endpoints

2026-09-05 16:05:17 +08:00

openai_grok_image_intent_gate_test.go

…

openai_image_intent_hint_test.go

…

openai_images_balance_test.go

fix(images): preserve insufficient balance failures

2026-09-21 10:23:10 +08:00

openai_images_controls_test.go

…

openai_images_failover_test.go

feat(openai): route OAuth image requests through native Codex Images

2026-09-11 15:57:09 +08:00

openai_images.go

fix(billing): hold inflight reservation until billing lands; price estimate like billing; wire remaining endpoints

2026-09-28 16:14:43 -04:00

openai_live_test.go

feat(composite): support Codex endpoints

2026-08-19 04:40:32 +00:00

openai_live.go

feat(composite): support Codex endpoints

2026-08-19 04:40:32 +00:00

openai_models_handler_test.go

fix(groups): merge main and resolve model allowlist conflicts

2026-09-07 18:33:39 +08:00

openai_models_handler.go

feat(gateway): retrieve models from the visible catalogue

2026-09-10 15:57:17 +08:00

openai_profit_slot_recheck_test.go

fix(openai): preserve remote compaction v2 responses endpoint

2026-08-14 16:23:52 +08:00

openai_profit_veto_budget_test.go

fix(profit-control): stop leaking profit policy, close veto livelock, restore passthrough turn pricing

2026-08-01 22:39:33 +08:00

openai_quota_platform_contract_test.go

…

openai_responses_astra_pro_failover_test.go

fix: preserve Astra reasoning mode for Codex OAuth

2026-09-07 17:58:20 +08:00

openai_responses_failover_cancel_test.go

…

openai_responses_image_intent_benchmark_test.go

…

openai_responses_input_tokens_test.go

修复 PR 5888 剩余审计问题

2026-08-20 22:01:20 -07:00

openai_service_tier_validation_test.go

feat(codex): support ultrafast service tier

2026-09-04 10:14:29 +08:00

openai_stream_validation_test.go

…

openai_stream_validation.go

…

openai_ws_client_close_attribution_test.go

fix(openai): 入站 WS 的客户端正常关闭与断开不再计为账号故障

2026-08-28 13:17:43 +08:00

openai_ws_turn_pricing_test.go

功能：支持渠道模型分时倍率定价

2026-08-17 19:45:07 +08:00

openai_ws_v2_passthrough_cyber_test.go

feat: add risk control user allowlist

2026-09-29 01:50:21 +08:00

openai_x_search_test.go

feat: Chat 与 Responses 往返保留 x_search，并补 sources 抽取

2026-08-13 08:37:57 +08:00

openai_x_search.go

feat: Chat 与 Responses 往返保留 x_search，并补 sources 抽取

2026-08-13 08:37:57 +08:00

ops_capture_writer_nil_test.go

修复 PR 5888 剩余审计问题

2026-08-20 22:01:20 -07:00

ops_error_logger_attribution_test.go

…

ops_error_logger_cyber_test.go

…

ops_error_logger_google_test.go

fix(ops): Gemini 原生路由的网关本地错误按错误类型归类

2026-10-10 23:37:48 +08:00

ops_error_logger_test.go

fix(gateway): 客户端断开统一标记 499,不再被记成 502/200 上游错误

2026-09-26 17:13:10 +08:00

ops_error_logger.go

fix(ops): Gemini 原生路由的网关本地错误按错误类型归类

2026-10-10 23:37:48 +08:00

ops_ingress_reject_capture_test.go

完善 OpenAI 网关切换与运维错误语义

2026-08-19 23:13:40 -07:00

ops_platform_test.go

…

page_handler_test.go

…

page_handler.go

…

passkey_handler_test.go

新增腾讯天御验证码认证门禁

2026-08-04 15:09:29 +08:00

passkey_handler.go

人机验证增加阿里云验证码 2.0

2026-08-04 20:57:15 +08:00

payment_handler_resume_test.go

…

payment_handler.go

feat(payment): 充值优惠阶梯支持折扣模式，快捷金额角标改为促销价签

2026-10-02 00:25:41 +08:00

payment_webhook_handler_test.go

…

payment_webhook_handler.go

…

redeem_handler_test.go

feat(redeem): paginate user redemption history

2026-09-16 13:19:15 +08:00

redeem_handler.go

feat(redeem): paginate user redemption history

2026-09-16 13:19:15 +08:00

request_body_limit_test.go

…

request_body_limit.go

…

request_body_parse_log_test.go

…

request_body_parse_log.go

…

request_body_read_log_test.go

fix(openai): harden Responses compatibility

2026-08-19 09:03:37 -07:00

request_body_read_log.go

fix(openai): harden Responses compatibility

2026-08-19 09:03:37 -07:00

security_audit_errors_test.go

…

security_audit_errors.go

fix(ops): Gemini 原生路由的网关本地错误按错误类型归类

2026-10-10 23:37:48 +08:00

security_audit_helper_test.go

fix(security-audit): restore websocket audit logs

2026-08-11 11:56:46 +08:00

security_audit_helper.go

fix(security-audit): restore websocket audit logs

2026-08-11 11:56:46 +08:00

security_audit_media_submit_test.go

…

security_audit_order_test.go

feat(gemini): 支持显式上下文缓存（AI Studio / Vertex / 自建网关上游）

2026-10-10 23:37:47 +08:00

seedance_test.go

feat: support native Seedance Ark video task API

2026-09-17 04:14:14 +09:00

seedance.go

feat: support native Seedance Ark video task API

2026-09-17 04:14:14 +09:00

setting_handler_public_test.go

fix: 完善腾讯验证码区域适配与 CSP 白名单

2026-08-06 20:34:37 +08:00

setting_handler.go

feat(settings): 新增「站点类型」三态开关，关闭订阅后用户端收起所有「订阅」入口与文案

2026-09-11 00:50:17 +08:00

stream_error_event_test.go

fix(responses): 为 grok-build 始终写出 sequence_number

2026-09-14 11:45:53 +08:00

stream_error_event.go

fix(responses): 为 grok-build 始终写出 sequence_number

2026-09-14 11:45:53 +08:00

subscription_handler.go

…

totp_handler.go

…

usage_handler_daily_test.go

…

usage_handler_request_type_test.go

feat(usage): track native compaction requests

2026-08-28 23:32:11 +08:00

usage_handler_sort_test.go

…

usage_handler.go

feat(usage): track native compaction requests

2026-08-28 23:32:11 +08:00

usage_record_submit_task_test.go

fix(grok): 三轮评审 — 流式 Search 去重与调度/计费加固

2026-08-07 17:58:23 +08:00

usage_record_task_fallback_test.go

fix(gateway): record observed usage when anthropic stream is interrupted

2026-08-01 11:29:58 +08:00

user_handler_test.go

fix(email): atomic verify-code attempts and hashed single-use reset tokens

2026-09-28 12:26:28 -04:00

user_handler.go

…

user_msg_queue_helper.go

完善 OpenAI 网关切换与运维错误语义

2026-08-19 23:13:40 -07:00

user_platform_quotas_handler_test.go

…

wire.go

feat(gemini): 支持显式上下文缓存（AI Studio / Vertex / 自建网关上游）

2026-10-10 23:37:47 +08:00