Commit Graph
9 Commits
Author SHA1 Message Date
Long Li 5a2f542ab0 fix(codex): prefer configured models over discovery
OpenAI groups can mix OAuth accounts with Responses-compatible API key accounts whose upstreams do not implement /models. The manifest handler previously selected one account and fetched its catalog before merging model_mapping aliases, so a permanent 404 returned 502 and hid models the administrator had already configured.\n\nTreat explicit model mappings from schedulable accounts as the authoritative Codex catalog. Build that catalog locally before account selection, preserve group filtering and final-body ETags, and retain the existing upstream discovery and failover path only when no mappings exist. Composite groups keep their existing account-first catalog behavior.
2026-08-22 14:43:40 +09:00
Long Li e471be7302 feat(codex): complete routed model catalogs
Codex treats the downloaded model catalog as a capability contract. Slug-only entries hide supported reasoning levels, can advertise invalid image inputs, and expose automatic or media-only models in the model picker. Generate complete route-aware metadata, preserve official manifests, filter non-agent models, and compute validators from the final catalog.
2026-08-22 14:43:40 +09:00
Kingsley 58e147fba6 feat(composite): support Codex endpoints 2026-08-19 04:40:32 +00:00
wucm667 3c62b5ca85 fix(openai): preserve web search for API-key Codex clients 2026-07-27 18:42:34 +08:00
superman2003andCursor 3ed2873f14 fix(openai): mark OAuth accounts unschedulable on Codex models manifest 401
The Codex models manifest path returned upstream 401s straight to the
client without feeding them into the account state machinery. A revoked
or invalidated OAuth account therefore stayed active and schedulable,
kept being selected for subsequent /models requests, and produced
repeated 502s until an admin ran a manual connection test (#4544).

- Route ChatGPT-backend manifest 401s through the shared upstream-error
  handling: token cache invalidation, temp-unschedulable cooldown for
  refreshable OAuth accounts, permanent disable for
  token_revoked/token_invalidated, plus the runtime scheduling block.
- Treat ChatGPT-backend manifest 401s as failover-eligible so the
  current /models request can switch to a healthy account instead of
  returning 502. Custom API key upstream 401s keep the existing
  no-failover, no-disable behavior since their /models auth is not
  authoritative for the account.
- Skip Agent Identity accounts: their 401s can be task-scoped and have
  a dedicated recovery flow.
- Attach the selected account to the ops error-log context so /models
  failures record the account_id.

Fixes #4544

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-07-19 21:13:27 +08:00
gebdalaoli-arch 3c68b2e369 fix: fail over Codex manifest accounts 2026-07-13 20:36:37 +08:00
gebdalaoli-arch ed31a52424 fix: stabilize API key Codex manifest refreshes 2026-07-13 19:31:38 +08:00
gebdalaoli-arch 0dce07ee8b fix: proxy Codex models through API key upstreams 2026-07-13 03:47:39 +08:00
limingeandClaude Fable 5 13e773ef5e feat: Codex 客户端模型清单(manifest)透传接口
背景:Codex CLI / Codex App 会从 provider 的 GET {base_url}/models?client_version=...
(自定义 provider 模式)或 GET /backend-api/codex/models(chatgpt_base_url 模式)
刷新模型选单,期望 ChatGPT Codex manifest 格式({"models":[{slug,...}]})。
sub2api 此前只提供 OpenAI 兼容格式的 /v1/models,Codex 客户端解析失败后静默
回落到本地缓存,导致指向 sub2api 的 Codex 客户端模型选单永久冻结在切换
provider 当天的状态,新模型(如 gpt-5.6 系列)永远不会出现。

方案:新增 manifest 透传——用组内可调度 OAuth 账号的凭据向
chatgpt.com/backend-api/codex/models 实时转发请求,响应体与 ETag 原样透传。
不在网关侧解析或维护模型清单:manifest schema 随 Codex 客户端版本演进,
透传保证网关无需跟进 schema 变化,且返回的始终是账号真实的模型权限
(区别于静态 DefaultModels 的"理论列表")。

路由:
- GET /backend-api/codex/models(新增)
- GET /v1/models 带 client_version 查询参数且组平台为 openai 时分发到
  manifest 透传(client_version 是 Codex 客户端的天然指纹,普通 OpenAI
  客户端不携带);其余请求保持原有 OpenAI 格式行为不变。

上游失败时按 fast-fail 返回错误,不伪造列表;Codex 客户端自身会回落缓存。

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-07-07 19:49:40 +08:00