github/sub2api
1
0
Fork 0
mirror of https://github.com/Wei-Shaw/sub2api.git synced 2026-10-07 15:08:02 +08:00
Code Issues Packages Projects Releases Wiki Activity
6,158 Commits 30 Branches 44 Tags
2f43e72bb95ab9984065b6ded35ab4cfe5541cdd
Commit Graph
4594 Commits
Author SHA1 Message Date
Wesley Liddick 2f43e72bb9 Merge pull request #6109 from feeeei/main 
feat(model-plaza): 模型广场增加长上下文阶梯计价显示 & 分时段计价显示
 2026-08-24 14:09:56 +08:00
Wesley Liddick b8651947c3 Merge pull request #6137 from yan9651688/codex/fix-cn-anthropic-usage-billing 
fix(billing): normalize CN Anthropic usage tokens
 2026-08-24 14:09:45 +08:00
yan9651688 695ebede70 fix(billing): normalize CN Anthropic usage tokens 2026-08-24 13:05:12 +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 625f1693cb Merge pull request #6118 from akihitohyh/fix/terminal-output-item-preservation 
fix(openai): rebuild streaming terminal output from the reported items
 2026-08-24 11:23:51 +08:00
Wesley Liddick f25f399be0 Merge pull request #5905 from wucm667/fix/issue-5883-restore-custom-tool-alias 
fix(openai): restore namespaced custom tool aliases
 2026-08-24 11:23:37 +08:00
Wesley Liddick 748b84a15a Merge pull request #6081 from wucm667/fix/issue-5942-deferred-tools 
fix(responses): remove orphan deferred tool flags
 2026-08-24 11:23:25 +08:00
Wesley Liddick fa42c3d706 Merge pull request #6080 from alfadb/fix/cc-stream-empty-tool-call-identity 
fix(openai): 剔除流式 tool_call 后续 delta 中的空 id/name
 2026-08-24 11:23:12 +08:00
Wesley Liddick fb01f5df2c Merge pull request #6060 from anguobao123/codex/document-openai-force-http-fallback 
fix(deploy): forward documented Gateway settings
 2026-08-24 11:22:38 +08:00
Wesley Liddick 8238956799 Merge pull request #6095 from xiaxiaxaia/fix/openai-oauth-upstream-model-sync 
fix(openai): sync models for OAuth accounts
 2026-08-24 11:21:58 +08:00
Wesley Liddick e00a8abdd5 Merge pull request #6124 from anguobao123/codex/diagnose-openai-load-batch-exclusions 
fix(scheduler): diagnose load-batch OpenAI exclusions
 2026-08-24 11:21:30 +08:00
Wesley Liddick a52665d079 Merge pull request #6061 from shunwang-crypto/fix/ops-mixing-cgroup-host-memory 
fix(ops): avoid mixing cgroup and host memory metrics
 2026-08-24 11:20:51 +08:00
feeeei b07d85c497 模型广场：分时计价同步渠道仅工作日规则 
- 阶梯表分时倍率透传渠道 weekdays_only；探针锚点显式固定在工作日
  （原 2026-01-01 恰为周四是巧合，锚点落周末会把仅工作日时段整组剔除）
- 前端时段徽章加「工作日」前缀，tooltip 说明周末全天按标准价计费
 2026-08-24 11:16:00 +08:00
alfadb cc894ef578 fix(openai): strip empty streamed tool-call id/name 
DashScope/DeepSeek later tool_call deltas send empty id and
function.name. Clients that merge with !== undefined overwrite
the first delta's identity and dispatch unknown tool "". Drop
those empty fields on the raw Chat Completions SSE path.
 2026-08-24 10:59:52 +08:00
feeeei 83d4eb6a43 模型广场：增加渠道分时段计价展示 
- 阶梯表查询附带分时倍率时段：时段取自计费解析到的渠道定价，
  每个时段的倍率由计费的 resolvedChannelTimeMultiplier 在时段内取值，
  分组价卡覆盖或配置非法时自然不出现；倍率为 1 的时段不列
- 广场模型条目新增 time_pricing（时区 + 时段 + 倍率）
- 前端把分时时段展开为独立行：模型名旁标注时段，价格按时段倍率折算，
  倍率列显示生效倍率；时区与计算口径放在提示中
 2026-08-24 10:50:52 +08:00
feeeei ecce0769c0 模型广场：上下文档位统一标签形态并保证升序 
- 阶梯表标签由计费层统一生成：有上限的档为「≤上限」、末档为「>下限」
  （达到阈值即进高档时用 < / ≥），不再沿用渠道区间的自定义 tier_label；
  合并同价段只看单价
- 前端档位按下限升序兜底展示，无标签时按同一形态生成
 2026-08-24 10:50:52 +08:00
feeeei 377d1230fc 模型广场：按计费阶梯单价表展示长上下文档位 
- 新建 ModelPlazaService（持计费服务与定价解析器）承接广场聚合，
  token 模型的单价与档位全部取自 ResolveContextPricingSchedule，
  渠道选择与计费同源；图片/按次模型沿用原档位合成
- 官方参考价改走计费目录（LiteLLM → 内置兜底 → 模型策略），带官方阶梯
- DTO 增加 long_context_pricing_enabled / long_context_basis /
  official_pricing.intervals
- 前端实付与官方三列按档分行（标签只在首列，其余列按行对齐），
  缓存列按档展示写/读价，边际计价以徽章与 tooltip 标注，
  分组关闭阶梯时在头部说明
 2026-08-24 10:50:52 +08:00
feeeei 6466978d2f 计费：统一 token 计费路径选择并提供上下文阶梯单价表查询 
- BillingService.CalculateTokenCostForRequest 承接网关的路径选择
  （分组/渠道定价 → 平台旧长上下文规则 → 内置目录），网关改为调用该入口
- Gemini /v1beta 的 200K 边际翻倍常量从 handler 移入
  BillingService.LegacyLongContextRule，入口只声明适用
- 新增 ResolveContextPricingSchedule：沿用 Resolve 解析链收集断点
  （渠道区间边界、目录阶梯阈值、旧规则阈值），每档单价由真实计费函数
  探针差商得出，倍率/策略变更无需同步；附阶梯表 vs 计费函数的对账测试
 2026-08-24 10:50:52 +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
Wesley Liddick 3e45d4e030 Merge pull request #6089 from lyen1688/feat/channel-time-pricing-weekdays 
新增渠道时间段定价工作日生效规则
 2026-08-24 10:21:47 +08:00
shaw 40aaf7b3ae fix: handle plugin route health update errors 2026-08-24 10:02:06 +08:00
shaw 684d9efb1f fix: harden plugin runtime and UI bridge 2026-08-24 09:50:46 +08:00
shaw 26ac0498f2 test: update plugin management settings contract 2026-08-24 09:16:16 +08:00
shaw 40ea3aebad feat: add OAuth outbound transport plugin system 2026-08-24 09:03:37 +08:00
anguobao123 3fd66a33be fix(scheduler): diagnose load-batch OpenAI exclusions 2026-08-24 01:25:54 +08:00
akihitohyhandClaude Opus 5 243921dc0a fix(openai): rebuild streaming terminal output from the reported items 
A terminal event that arrives with an empty output was rebuilt from delta
accumulation. BufferedResponseAccumulator models only one reasoning item, one
message, and N function calls, and records no item id, status, or phase, so a
turn carrying several items collapsed into a single fabricated message: the
reasoning item disappeared, the real message id was replaced, and phase was
lost.

reconstructResponseOutputFromSSE already prefers the raw output_item.done
items over accumulation for buffered responses. The streaming path had no
equivalent because it never sees the whole body at once. Collect the raw item
of each output_item.done keyed by output_index and rebuild from those, falling
back to accumulation only when the stream reported no done item at all.

Items are stored as raw JSON, so vendor extensions and item types this gateway
does not model survive the rebuild verbatim.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
 2026-08-23 20:49:36 +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 4eadee1074 [verified] test(openai): update responses bridge signature 2026-08-23 08:45:04 +08:00
wucm667 31d5b67baa fix(openai): restore namespaced custom tool aliases 2026-08-23 08:11:20 +08:00
xiaxiaxaia 913ec5d74b fix(openai): sync models for OAuth accounts 2026-08-23 02:07:59 +08:00
lyen1688 77e0409f7c 新增渠道时间段定价工作日规则 2026-08-23 00:30:27 +08:00
wucm667 7498d8fdc8 fix(openai): enforce serial tool calls for Responses Lite 2026-08-22 19:40:19 +08:00
shunwang-cryptoandClaude Opus 4.8 cd05772e91 fix(ops): avoid mixing cgroup and host memory metrics 
In Docker + cgroup v2 with no memory limit set, /sys/fs/cgroup/memory.current
returns a small container number while /sys/fs/cgroup/memory.max is "max".
readCgroupMemoryBytes then returned (used=<container>, total=0, ok=true).

collectSystemStats used that container "used" but, being unable to derive a
cgroup total, filled the total from the host via gopsutil. The dashboard then
computed container_used / host_total, e.g. ~60MB / 23GB ≈ 0.3% — wildly
understating real usage.

Fix: introduce resolveMemoryStats, which picks a single self-consistent
(used, total, percent) trio from ONE source. cgroup metrics are used only when
the cgroup exposes both a current usage AND a concrete limit (memory.max != max,
so total > 0); otherwise used/total/percent all fall back to the host reading.
The two sources are never mixed.

- memory.current valid + memory.max = "max"  -> all host metrics
- memory.current = 512MiB + memory.max = 2GiB -> ~25% from cgroup
- no cgroup (bare metal)                       -> all host metrics

CPU metric behavior is unchanged (cgroup attempt then host fallback).

Adds ops_metrics_collector_memory_test.go covering all branches.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 2026-08-22 19:35:31 +08:00
wucm667 7a09a2eaf7 fix(responses): remove orphan deferred tool flags 2026-08-22 16:42:56 +08:00
Wesley Liddick d45135d87d Merge pull request #6068 from okbexx/fix/codex-guardian-parent-affinity 
fix(openai): keep auto-review on parent account
 2026-08-22 13:41:42 +08:00
Wesley Liddick d29d7f8cbc Merge pull request #6065 from chinnsenn/fix/image-generation-flows 
fix(openai): stabilize oauth image generation
 2026-08-22 13:35:26 +08:00
Wesley Liddick fd6cd474d6 Merge pull request #5846 from lbyxiaolizi/fix/responses-chat-malformed-tool-arguments 
fix(apicompat): reject malformed tool-call arguments
 2026-08-22 13:35:02 +08:00
Wesley Liddick 73f6a590bf Merge pull request #5912 from xuhaihan/fix/deepseek-responses-client-tools 
fix(deepseek): adapt Codex client tools for Responses
 2026-08-22 13:34:49 +08:00
Wesley Liddick ffc01f9c66 Merge pull request #5864 from wucm667/fix/issue-5850-http-bridge-replay 
fix(openai): avoid duplicate HTTP bridge replay
 2026-08-22 13:34:32 +08:00
Wesley Liddick 6244090c1c Merge pull request #5487 from an-epiphany/fix/file-part-min 
fix(apicompat): chat/completions 的 file part 转换为 Responses input_file，不再静默丢弃
 2026-08-22 13:34:18 +08:00
Wesley Liddick 844b118785 Merge pull request #5938 from Hakunm/fix/google-one-model-catalog 
fix(gemini): 限制 Google One OAuth 模型目录 / constrain Google One model catalog
 2026-08-22 13:34:05 +08:00
alfadb 86470628df feat(ollama): 对 Ollama Cloud 账号 clamp max_tokens 上限 2026-08-22 12:38:05 +08:00
Jarl fa4587041c fix(openai): keep auto-review on parent account 2026-08-22 12:14:19 +08:00
chinnsenn cb8dabc12f fix(openai): stabilize oauth image generation 2026-08-22 11:21:39 +09:00
alfadb b30651a0ad fix(ollama): 对齐 Cloud Chat Completions 思维字段为 reasoning_content 
Ollama Cloud 的 raw /v1/chat/completions 把思维放在 reasoning/thinking，
DeepSeek 客户端只认 reasoning_content。仅对 openai+apikey+force_chat_completions
且具备 Ollama Cloud 信号的账号，在 raw CC 直转路径做 wire JSON 双向补齐。
 2026-08-22 09:26:23 +08:00
xuhaihan cef18b4ad3 fix(deepseek): route client tools through native responses 2026-08-22 01:04:11 +08:00
xuhaihan 30ae15268f Merge remote-tracking branch 'origin2/main' into fix/deepseek-responses-client-tools 2026-08-22 00:52:33 +08:00
anguobao123 9f2f2738fd docs(openai): document force HTTP fallback 2026-08-21 23:43:09 +08:00
Wesley Liddick 67380eafd5 Merge pull request #5549 from zcxads666/fix/openai-capabilities-empty-set 
fix(openai): 允许空 openai_capabilities 的 OAuth 账号加入文本调度
 2026-08-21 21:53:27 +08:00
Wesley Liddick 2ddda67354 Merge pull request #6049 from MokoYee/fix/openai-sticky-prefix-system 
修复 OpenAI Chat 动态系统消息导致账号亲和漂移
 2026-08-21 21:34:57 +08:00