Commit Graph
67 Commits
Author SHA1 Message Date
shaw 095b525360 fix(openai): pin Responses Lite parallel tool calls 2026-08-25 12:14:52 +08:00
Jiao Ziang 53d76ad800 fix(openai): enforce Responses Lite tool call mode 2026-08-25 00:21:29 +08:00
Wesley Liddick 7bb9c0ed7d Merge pull request #6079 from okbexx/fix/codex-analytics-account-affinity
fix(openai): scope Codex identity to OAuth account
2026-08-24 14:54:05 +08:00
Wesley Liddick 5f43696a9a Merge pull request #6121 from creamtea47/codex/feat-openai-auto-reset-credit
feat: OpenAI 重置卡按用量阈值自动使用
2026-08-24 14:39:59 +08:00
NellPoi 6f972145b7 feat: 支持 OpenAI 重置卡按用量阈值自动使用 2026-08-24 13:28:33 +08:00
alfadb f06bf181d2 feat(openai): support Fast mode service_tier across responses/chat/WS paths
- Accept fast|priority (canonical priority), flex|auto|default|scale on
  /v1/responses and /v1/chat/completions; reject unknown/empty/non-string
  with HTTP 400; omitted and null stay compatible.
- Propagate service_tier through JSON/SSE, Responses<->Chat conversions,
  fallback paths and HTTP->upstream WebSocket bridge.
- Billing prefers the upstream terminal tier; the outbound (policy-
  transformed) tier is used only when upstream omits the field.
  Explicit upstream default bills Standard even when Fast was requested.
- Pricing: Fast premium 2x Standard for gpt-5.6-sol/terra/luna and
  gpt-5.4; 2.5x for gpt-5.5; channel FastMultiplier stays authoritative.
- Live verification (official Codex 0.149.0 + gateway, HTTP & WS):
  upstream ChatGPT backend may return terminal default even when the
  account catalog advertises priority; billing follows the actual tier.
2026-08-24 11:52:48 +08:00
Wesley Liddick c416467882 Merge pull request #6084 from wucm667/fix/issue-6057-responses-lite-parallel-tools
fix(openai): enforce serial tool calls for Responses Lite
2026-08-24 11:24:17 +08:00
Wesley Liddick 3b8a148bcf Merge pull request #6111 from feeeei/fix/request_billing
fix(billing): bill fast mode by the tier upstream actually served
2026-08-24 10:43:17 +08:00
shaw 40ea3aebad feat: add OAuth outbound transport plugin system 2026-08-24 09:03:37 +08:00
feeeei 75faedda95 计费:fast/priority 按上游响应实际档位只降不升计费
此前 OpenAI service_tier 与 Anthropic speed 的计费档位只看请求侧,上游
因容量把 priority/fast 静默降为 default/standard 时仍按 2x 收费。现在:

- upstreamResponseModelObserver 同时观察响应侧档位:OpenAI 只信终止事件
  与无类型 body 里的 service_tier(response.created 为请求回显,忽略),
  Anthropic 读 usage.speed;查找仅在带 model 的帧上触发
- ResolveBillingServiceTier 只降不升:响应档位比请求便宜才采纳,更贵、
  未知或缺失一律沿用请求侧,与图片尺寸归并、response_model 计费的
  "不得更贵"不变式一致
- usage_logs.service_tier 记录实际计费档位,降级写
  billing.service_tier_downgraded 审计日志;HTTP、WS 及 ws_v2 relay 全覆盖
2026-08-23 16:53:28 +08:00
wucm667 7498d8fdc8 fix(openai): enforce serial tool calls for Responses Lite 2026-08-22 19:40:19 +08:00
Jarl d493ce0bb2 fix(openai): scope codex identity to oauth account 2026-08-22 16:06:04 +08:00
xuhaihan cef18b4ad3 fix(deepseek): route client tools through native responses 2026-08-22 01:04:11 +08:00
IanShaw b2b2adcf8d 修复 PR 5888 审查发现的兼容性与竞态问题 2026-08-20 21:06:48 -07:00
IanShaw 48615d5d1f Merge remote-tracking branch 'upstream/main' into fix/openai-responses-compatibility
# Conflicts:
#	backend/internal/handler/ops_error_logger.go
2026-08-20 00:31:12 -07:00
IanShaw acce29af27 补齐 OpenAI 与 Grok 协议兼容处理 2026-08-19 23:13:29 -07:00
SavitarC b3092145db fix(accounts): harden adaptive protocol compatibility 2026-08-20 13:57:25 +08:00
SavitarC 85051616f9 feat(accounts): add adaptive API protocol routing 2026-08-20 11:52:05 +08:00
IanShaw cf3577a3c5 fix(openai): harden Responses compatibility 2026-08-19 09:03:37 -07:00
Wesley Liddick 58ea46e894 Merge pull request #5661 from wucm667/fix/issue-5659-openai-custom-tools
fix(openai): restore API-key custom tool calls
2026-08-18 15:47:50 +08:00
spongehah bb6c3b4f6a fix: unify Codex OAuth outbound identity onto the inference resolver
Token exchange, PAT whoami, models, probes, and pre-writes now follow
the same UA/version chain as Codex inference instead of hardcoded
codex-cli/0.91.0 or compile-time constants.
2026-08-18 13:03:22 +08:00
Jarl 6793d5ac85 fix(openai): make Codex convergence identity consistent 2026-08-17 20:25:28 +08:00
Randark 901a0439f1 feat: 国产供应商一等支持(Kimi/Zhipu/DeepSeek 多协议 + 配额/余额监控)
后端:
- 协议凭证维度 credentials[api_protocol] ∈ chat_completions(默认)/anthropic/responses(deepseek)
- /v1/messages 零转换直通原生 Anthropic 端点(kimi/zhipu/deepseek),CC/Responses
  入站交叉组合走 apicompat 双向转换链(responses/chat_completions anthropic-native 转发器)
- count_tokens:anthropic 协议透传原生端点;其余 CN 协议本地 tiktoken 估算
- Coding Plan 额度探测(5h/weekly 滚动窗口)+ payg 余额探测(kimi/deepseek),
  deepseek 双币种 CNY+USD 明细,任一币种达标不停调
- 周期任务 [CNBalance] 并发探测 + 预算随工作量放大;响应式 429 冷却到最早窗口
  重置点;余额不足可恢复临时停调;智谱 CREDIT_LIMIT 不污染窗口解析
- CC→anthropic 流式客户端断开后继续排水上游保住 usage 计量

前端:
- 创建/编辑弹窗 account_mode + api_protocol + base_url 联动预设(含 watcher 竞态防护)
- 用量单元格:kimi/zhipu coding 显示 5h/weekly 窗口,kimi/deepseek payg 显示余额,
  多币种并列展示;探测失败保留快照;挂载自动探测 5min 去抖
- 调度阈值设置面板补 kimi/zhipu 平台(对齐后端 AllowedSchedulingThresholdPlatforms)
2026-08-15 10:37:51 +00:00
shaw fce41e318f fix(openai): make Codex fingerprint convergence opt-in and cover passthrough
Default codex_fingerprint_mode to off. v0.1.175 treated a missing key as
"session", so upgrading silently rewrote installation/session/thread/turn/
window identifiers for every existing OAuth account that had never configured
this field. The quota regressions in #5555, #5556 and #5582 line up with that
version boundary, with A/B reports that rolling back to v0.1.173 restores
quota. Convergence is now explicit opt-in (#5610).

Only accounts that never set the field change behaviour; explicit off /
device / session / full keep working exactly as configured. That required
flipping the persistence condition in all three account modals from
"!== 'session'" to "!== 'off'": the old rule deleted the key when it equalled
the default, which after the flip would have silently discarded an
administrator's explicit opt-in to session.

Also extend convergence to the passthrough path, which previously left client
identifiers untouched:

- resolve the ids once in forwardOpenAIPassthrough and rewrite
  client_metadata on the raw bytes (gjson extract + sjson splice) because
  passthrough is a hot path that must not fully unmarshal multi-MB bodies;
  a shared core keeps the raw and map variants from drifting
- both request builders apply the staged ids at the same relative position
  (after session isolation, before identity enforcement) so headers and body
  share one id set and turn_id stays consistent
- stage the ids unconditionally, including nil: a failover from a converged
  account to an off account must not leave the previous account's ids behind
2026-08-15 16:35:26 +08:00
wucm667 44ef88f659 fix(openai): restore API-key custom tools 2026-08-15 11:02:59 +08:00
InCerryGit 9662cff2e7 fix(openai): preserve remote compaction v2 responses endpoint 2026-08-14 16:23:52 +08:00
shaw c0ab3a00ea feat: Codex OAuth 设备指纹收敛,减少上游可见的设备数和会话数
多人共享同一 OAuth 账号时,各用户 Codex 客户端携带各自不同的 installation_id/session_id/thread_id,
上游据此判定设备数和会话数并限制配额。本功能将这些标识改写为账号级恒定值。

四档策略(账号级 extra 字段 codex_fingerprint_mode):
- off: 不做任何收敛,原样透传
- device: 仅收敛 installation_id
- session(默认): 收敛 installation_id + session_id,thread_id 按客户端原始 session 派生
- full: 收敛所有标识(installation_id + session_id + thread_id)

改写覆盖 6 个指纹载体:x-codex-turn-metadata 头(JSON 内部字段)、x-codex-window-id、
x-codex-installation-id、x-client-request-id、session-id/session_id/thread-id、
请求体 client_metadata。头和体共享同一份预计算 IDs 确保 turn_id 等随机字段一致。
2026-08-12 17:57:50 +08:00
IanShaw027 1f58e25ab3 Merge upstream/main into feat/grok-complete-integration
冲突集中在 chat completions / messages 两条 Responses 转发路径:
upstream 给 OpenAIForwardResult 增加了 UpstreamResponseModel 与
UpstreamResponseModelConflict(配套 beginUpstreamResponseModelObservation
观测器),本分支在同样位置把返回值改成了具名变量以便挂 Grok 原生搜索计数。
两侧不互斥,合并结果同时保留上游的响应模型观测字段与 Grok SearchCount 逻辑。

frontend/pnpm-lock.yaml 取 upstream 版本:package.json 与 upstream 完全一致,
本地差异只是 pnpm install 的重解析噪音。
2026-08-08 11:12:56 +08:00
Wesley Liddick cc67b1aca1 Merge pull request #5406 from bestony/fix/openai-oauth-routing-hints
fix(openai): forward OAuth routing hints
2026-08-08 10:58:53 +08:00
白宦成 de349187d9 fix(openai): harden priority routing hints 2026-08-07 21:47:54 +08:00
白宦成 815035fcc9 fix(openai): send OAuth routing hints 2026-08-07 21:47:53 +08:00
白宦成 915cc7e7bd fix(openai): stop injecting legacy beta on OAuth responses 2026-08-07 21:47:53 +08:00
Brisbanehuang db0bff82c7 feat(usage): audit upstream response models
(cherry picked from commit 839036224f795c8ee5dc6718a2a14372a45eea44)
2026-08-07 09:40:11 -04:00
li f3c94d2099 fix(openai): 修正 Responses 工具 Schema 中显式为 null 的 parameters.type
Codex Desktop 内置的 automation_update 工具会带 parameters.type = null,
OpenAI 直接回 400 invalid_function_parameters,而网关把该状态归一成可重试的
502 upstream_error。该工具定义还会沉进多轮会话历史,之后每一轮都继续失败,
同一份坏 Schema 在账号池里被反复重放。

在 OpenAIGatewayService.Forward 分流到 passthrough / Codex transform /
原生 ChatCompletions 之前,统一把显式为 null 的 parameters.type 补成
"object",覆盖顶层 tools[] 与多轮历史里 input[].tools[] 的嵌套工具定义。

只修正显式 null:缺失 type 的 Schema 本身合法,补写会收窄客户端语义,
因此保持原样。

Fixes #5364
2026-08-07 20:30:44 +08:00
IanShaw027 ccf7ba3ba7 fix(grok): P2 分层清理、主对话 SearchCount 与调度 7d/30d
- DoGrokNativeResponsesJSON 去 gin.Context,UA 固定 CLI 身份
- GrokOAuthService 去掉 redis 直依赖,session store 由 wire 注入
- 主对话/流式响应统计 web_search_call/x_search/tool_search 写入 SearchCount
- Grok 调度阈值候选并入官方 weekly/monthly billing %
2026-08-07 17:21:01 +08:00
zhiyu 2eb24814fe fix(codex): 强制统一出站身份并让客户端版本号跟随官方发布
上游 /backend-api/codex 在容量紧张时按客户端身份分优先级降载,被降载的请求
HTTP 200 后立刻推流内 server_is_overloaded。此前网关对配不出官方身份的客户端
整体回退到硬编码的 codex_cli_rs/0.144.1(落后官方 4 个发布),这些请求稳定
落在被优先丢弃的一侧。

- 强制统一出口:所有 OAuth 出站的 User-Agent / originator / version 一律改写
  为网关规范身份,客户端自报身份不参与构造;HTTP / 透传 / WS / alpha-search /
  探针全覆盖。compat 桥接故意删除 originator 的路径保持 no-op。
- 版本号收敛为单一来源,运行时优先级为面板覆写 → 自动同步值 → 内置常量;
  UA 与 version 头同源派生,不再各自硬编码。
- 新增 3 小时自动同步官方客户端最新稳定版,面板可关闭,无需为跟版本而发版。
- 流内 server_is_overloaded / slow_down 改为先在同账号有界重试再切号,并标记为
  请求级瞬时故障,不再据此临时封禁账号。
- 移除被取代的降载身份黑名单、浏览器 UA 兜底及其辅助函数。
2026-08-03 20:14:58 +08:00
InkCrow 272735b0a7 fix(openai): preserve Codex namespace tools on OAuth Responses forwarding
Flattening Codex `type:"namespace"` tool declarations into `namespace__name`
renames the tool but cannot rewrite the contract the client already handed the
model: Codex tells it to call `to=functions.collaboration.spawn_agent`. Under
`tool_mode: code_mode_only` (gpt-5.6-*) collaboration tools are the model's only
direct channel, so the rename left it unable to address them — it fell back to
`functions.exec` and looped on meaningless shell commands until the session
collapsed, with no upstream error anywhere. Fixes #4978.

Preserving is what the upstream actually wants. OAuth egress is unconditionally
chatgpt.com/backend-api/codex/responses (base_url is read only for API-key
accounts), i.e. the party that defines the extension; codex-rs builds one
request for both WS and HTTP and falls back WS->HTTP mid-session, so these
declarations already reach that host over HTTP — as they do today via the
responses-lite carrier and the WS/HTTP bridge, neither of which ever flattened.

Namespace names cannot be allowlisted: `features.multi_agent_v2.tool_namespace`
is user-configurable (openai/codex#31864 recommends renaming it to `agents`) and
MCP/connector namespaces are generated at runtime (mcp__codex_apps__gmail).
So the default is inverted rather than extended with more names.

- Preserve namespace declarations by default for OpenAI OAuth; the new
  per-account `extra.openai_responses_flatten_namespaces` restores the previous
  behavior for deployments routing OAuth traffic to a relay that rejects them.
- Keep `input[].namespace` on tool-call items for OAuth non-compact requests.
  The upstream requires the round-trip ("Missing namespace for function_call
  '...'. Round-trip the model's function_call item with its namespace field
  included."), so preserving declarations while stripping the calls would 400 on
  the second turn. The item-type allowlist mirrors the existing reactive strip.
- Keep compact on its current behavior: that endpoint rejects the field outright
  ("Unknown parameter: 'input[894].namespace'", #4761) and there is no evidence
  either way about namespace declarations there, so this change does not widen
  its surface. API-key accounts are untouched — their upstream is a standard
  Responses API and the reactive retry only clears one index per round trip.
- Clear the flatten mapping at the start of each Forward attempt so an account
  reached through failover cannot restore responses with the previous account's
  mapping.

Admin UI gets an OAuth-only toggle (create/edit/bulk) plus zh/en copy noting the
compact exception.
2026-07-28 09:19:13 +08:00
Wesley Liddick 333acde7d1 Merge pull request #4832 from ListenCodes/fix/openai-apikey-responses-item-id-v2
fix(openai): sanitize API-key responses item IDs
2026-07-25 13:42:35 +08:00
ListenCodes c5d9d57940 fix(openai): sanitize API-key responses item IDs 2026-07-24 20:15:58 +08:00
scp-planet 1bdf99109b fix(openai): strip input namespaces for api key accounts 2026-07-24 02:59:32 +08:00
scp-planet e70fbf3204 fix(openai): strip residual input namespaces before HTTP forwarding 2026-07-24 02:59:32 +08:00
superman2003 f8445a241b fix(grok): round-trip Codex client tools on Responses 2026-07-22 12:15:48 +08:00
haruka 1f7b7b9132 fix: honor temp unschedulable rules in pool mode 2026-07-17 18:12:33 +08:00
harukaandClaude Opus 4.8 605b026cc4 fix(gateway): route image-intent /v1/responses only to Responses-capable accounts (#4417)
For OpenAI-compatible API-key accounts, /v1/responses requests with
image-generation intent could be scheduled to accounts whose upstream
does not support the Responses API (extra.openai_responses_supported=false).
The flag was only consulted at forward time, where such accounts are
silently downgraded to a Chat-Completions path that cannot produce images,
causing upstream 4xx/5xx or canceled requests.

Fix:
- Add endpoint capability OpenAIEndpointCapabilityResponses. Its check in
  SupportsOpenAIEndpointCapability excludes only OpenAI API-key accounts
  probed as unsupported (mirroring the forward-time downgrade condition);
  OAuth/Grok/unprobed accounts keep existing behavior, and a responses-
  capable upstream must still pass the chat_completions gate. Reusing the
  existing requiredCapability plumbing makes every scheduler filter path
  enforce it with no scheduler signature changes.
- Request the responses capability at the HTTP Responses and
  ResponsesWebSocket call sites only when imageIntent && platform==openai,
  so non-image requests keep the downgrade path and Grok's own image path
  is untouched.
- Normalize max_tokens -> max_output_tokens on the native responses
  forward path (PlatformOpenAI), and strip prompt_cache_options alongside
  prompt_cache_retention/safety_identifier.

/v1/images/generations continues to use native image capability (unchanged).

Tests: capability truth table, scheduler exclusion of unsupported accounts,
and forward-path transform behavior.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KXpzKKvsb5jW2GvgBQqnnZ
2026-07-16 00:23:40 -07:00
Wesley Liddick 51f2b0e3ee Merge pull request #4382 from wp-a/fix/openai-responses-rejected-field-retries
[codex] retry explicitly rejected Responses fields
2026-07-16 09:37:36 +08:00
王鹏 3db00d3fee fix(openai): fail over account-specific body limits 2026-07-16 00:19:20 +08:00
王鹏 5e4da92dec fix(openai): retry explicitly rejected Responses fields 2026-07-15 22:37:30 +08:00
Wesley Liddick 8e84071f21 Merge pull request #4307 from wp-a/fix/openai-first-output-timeout
fix(openai): bound native responses first output wait
2026-07-15 14:41:11 +08:00
王鹏 fc4089f292 fix(openai): bound native responses first output wait
Add an opt-in first semantic output budget for native HTTP Responses, including response-header wait. Keep preamble and keepalive bytes non-semantic so a stalled account can fail over once without replaying its response IDs. Defaults remain disabled.

Related to #4201, #4185, and #4248. Complements the HTTP/2 dead-connection fix in #4207.
2026-07-15 13:01:16 +08:00
superman2003 bc6d9c47d2 fix(openai): normalize Responses Lite tool declarations 2026-07-15 11:37:38 +08:00