Commit Graph
4923 Commits
Author SHA1 Message Date
qiujunandClaude Opus 5 e7c029875c test(billing): 图片输出按 output 子集计价
上一条改用统一定价管线后,TestTryModelFilePricing_WithImageOutput 挂在 0.3 vs
0.28:原算式把 ImageOutputTokens 当 OutputTokens 之外的额外量叠加,等于按文本
output 价和图片价各收一次,测试固化了这个重复计费。

ImageOutputTokens 本就是 OutputTokens 的子集(Gemini 取自 candidatesTokensDetails
的 IMAGE 档,OpenAI 取自 output_tokens_details),统一管线先扣除再按图片单价计,
未配图片价时回退到 output 价、总额不变,正是这个语义。期望改为 0.28。

影响:涉及图片输出的模型,账号统计成本会降低"图片 token × 文本 output 单价"。

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-31 12:33:06 +08:00
Wesley Liddick 0678b24d58 Merge pull request #6283 from okbexx/feat/usage-compaction-kind
feat(usage): track native compaction requests
2026-08-31 11:19:37 +08:00
Wesley Liddick f34735f462 Merge pull request #6388 from alfadb/feature/ollama-usage-cn-platforms
feat(ollama): 用量窗口支持挂在国产三家平台下的 Ollama Cloud 账号
2026-08-31 11:16:25 +08:00
Wesley Liddick a2d7d41185 Merge pull request #6325 from 251392463/fix/grok-responses-vision-tool-output
fix(grok): preserve Responses vision tool output images
2026-08-31 11:15:33 +08:00
Wesley Liddick 03ab687686 Merge pull request #5866 from wucm667/fix/issue-5863-scope-monitor-groups
fix: scope channel monitor groups by user access
2026-08-31 11:15:13 +08:00
Wesley Liddick fbc69322af Merge pull request #6254 from wucm667/fix/issue-6249-configurable-image-cooldown
feat(openai): configure image tool cooldown
2026-08-31 11:13:52 +08:00
Wesley Liddick 414489d152 Merge pull request #5469 from wucm667/fix/issue-5467-passthrough-http-bridge
fix(openai-ws): bridge oversized passthrough requests
2026-08-31 11:13:35 +08:00
Wesley Liddick 50a9dd7a64 Merge pull request #6386 from wucm667/fix/issue-6383-isolate-passthrough-ws
fix(openai): isolate passthrough websocket sessions
2026-08-31 11:13:16 +08:00
Wesley Liddick a4156eea10 Merge pull request #6385 from wucm667/fix/issue-6381-preserve-quota-limits
fix(admin): preserve quota limits on partial group updates
2026-08-31 11:12:56 +08:00
wucm667 897faea333 fix(account): reset quota cooldown atomically 2026-08-31 09:12:09 +08:00
feeeei 0e08c39999 fix(openai-ws): remember rejected encrypted content across turns
invalid_encrypted_content 恢复此前只清理当次请求体:同一失效密文随客户
端维护的会话历史在后续每一轮重新出现,反复触发"整包被拒→剥离→重试/
重连"(#6382 生产观测 2h 内 251 次 recovery、273 次 WS 重连),既是
QoS 放大器也是内存压力放大器。

按会话记录被上游拒绝过的 encrypted_content SHA-256 摘要(lineage):
- OpenAIWSStateStore 新增会话级摘要绑定,TTL 复用 session sticky,
  单会话 512 摘要为存储自保护,超出仅退化为既有单次 recovery
- 写点覆盖 forward WSv2 recovery、forward 非 WS 400 重试、ingress
  error 事件、bridge 400/流内 error;读点在 forward 进场与 ingress/
  bridge turn 循环入口按摘要预剥离,新密文摘要不同不受影响
- 只收剥离端可处理的项类型(reasoning/compaction/compaction_summary),
  避免记入永远剥不掉的摘要后每轮空转解码
- replay 历史序列同步剥离,避免与已剥离的当前 input 项错位导致
  prefix 复用失配

含端到端集成测试:turn1 上游拒绝即记录,turn2 上游收到的请求不再携带
失效密文且 reasoning 骨架保留。
2026-08-31 01:02:13 +08:00
feeeei ca9b4d73f2 fix(openai-ws): share immutable replay bodies to stop O(T²) copying
OpenAI Responses WS/HTTP-bridge 的 replay 历史此前每轮对全部正文做多次
深拷贝(构建、双份序列、保存点各一遍),长会话累计分配 O(T²),实测
128 turn x 10KiB 达 160MB+,是 #6382 内存持续增长直至 OOM 的主因。

改为共享不可变正文所有权:
- extract 经零拷贝 string 视图 + gjson Index 直接切 payload 子切片,
  并以 json.Valid 保持对断裂数组的既有错误语义
- 序列构建/sanitize/collector/保存点全部只传引用;跨持有者保存经
  combineOpenAIWSReplayItems 新建头数组,正文不复制
- bridge 每轮一次解析,正常 replay 与 account-failover 序列共享正文
- 前缀比较增加字节级快路径,避免逐轮全历史 decode+encode
- strict 状态可用时不再整包保留上一轮 payload

新增 128 turn 分配回归测试锚定上界(实测 2.5MB,深拷贝回归会击穿),
以及正文共享的所有权断言。
2026-08-31 01:01:20 +08:00
qiujunandClaude Opus 5 9eabd2a5b7 fix(billing): apply model pricing policies to account stats cost
resolveAccountStatsCost 的优先级 3 维护了第二份"单价 × token 数"实现,每加一个
定价特性都要手工镜像:service tier 与长上下文分别由 9261dd77、25e6688a 补齐,
而 b5827cfd 引入的 DeepSeek 官方峰谷一直没有生效,高峰时段账号成本最多低估一半。

改为复用 CalculateCostWithServiceTier。channelPricing 仍为 nil,优先级 3 的语义
不变:只取模型定价文件、不引入渠道自定义定价;低谷时段结果与原算式完全一致。

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-30 22:57:53 +08:00
wucm667 d5a0124637 fix(openai): isolate passthrough websocket sessions 2026-08-30 16:49:01 +08:00
alfadb 30b29e51e8 feat(ollama): 用量窗口支持挂在国产三家平台下的 Ollama Cloud 账号
Ollama Cloud 用量窗口(浏览器会话、5h/7d 快照、自动刷新、同 key 组共享)此前
只认 platform ∈ {openai, anthropic} 的 apikey 账号。但网关对 kimi/zhipu/deepseek
的 apikey 账号同样按 credentials.base_url 转发,用户把 Ollama Cloud 的 key 挂在
这三个平台下跑 Ollama Cloud 托管的 glm/kimi/deepseek 模型时,请求转发正常,用量
窗口却完全不可用。

- 平台白名单放开到 {openai, anthropic, kimi, zhipu, deepseek};type='apikey'
  与官方 ollama.com base_url 的限制不变
- Go 侧收敛为 isOllamaCloudUsagePlatform,SQL 侧收敛为
  ollamaCloudUsagePlatformsSQL,6 处白名单只引用单一来源,并补两侧互为镜像的
  漂移守卫测试
- CN 周期余额/额度探测跳过 ollama 用量账号:其探测端点由 base_url 衍生,对
  ollama.com 会被出站 URL 白名单拒绝,每轮都白跑并产生噪声
- 前端账号列表 CN 分支在 eligible 时改渲染 Ollama 用量窗口,并跳过 CN 额度/
  余额子单元格(同上,渲染出来只会是一行探测报错)
2026-08-30 16:38:11 +08:00
wucm667 9f1effd71c fix(admin): preserve quota limits on partial group updates 2026-08-30 16:36:13 +08:00
Long Li 57c76584a6 fix(codex): advertise priority tier for fast models 2026-08-30 14:21:08 +09:00
Ritel-T 82105f260a fix(billing): resolve shadow credential tier contract 2026-08-30 13:40:37 +09:00
Ritel-T 50ad6e2e50 fix(billing): narrow Codex OAuth default exception 2026-08-30 13:40:19 +09:00
Ritel-T f323d84644 test(billing): cover Codex OAuth usage pricing 2026-08-30 13:07:33 +09:00
Ritel-T 3a9070359b fix(billing): scope Codex OAuth tier trust 2026-08-30 13:07:21 +09:00
Ritel-T d39fc491ec test(openai): cover websocket tier separation 2026-08-30 13:07:07 +09:00
Ritel-T 2b8cb628b6 fix(openai): preserve tiers in messages chat fallback 2026-08-30 13:06:45 +09:00
Ritel-T 8b4b3f4a94 fix(openai): preserve tiers in responses chat fallback 2026-08-30 13:06:32 +09:00
Ritel-T 624e4eef64 fix(openai): separate outbound and observed tiers 2026-08-30 13:06:11 +09:00
wucm667 a3bbf33c05 fix(monitor): scope groups by user access 2026-08-30 09:28:40 +08:00
wucm667 7c616db07a fix(openai-ws): bridge oversized passthrough requests 2026-08-30 08:41:08 +08:00
alfadb b5827cfd54 fix(pricing): align DeepSeek billing with official peak/off-peak rates
同步 DeepSeek 官方峰谷价格与模型定价。
2026-08-29 17:30:23 +08:00
m11y ea4291a92c fix(anthropic): scope Fable scheduling threshold to Fable models
让 Fable 7d_oi 调度阈值仅限制 Fable 模型族,不暂停整个 Anthropic 账号。
2026-08-29 17:30:18 +08:00
101Harry 32ac921f24 fix(anthropic): avoid Fable OAuth system prompt refusal
修复 Fable OAuth 请求因系统提示词形态导致空响应的问题。
2026-08-29 17:29:30 +08:00
Ryan Chou eb4237a2b0 fix(billing): 渠道定价对带后缀的模型名不再被官方兜底价覆盖
修复带后缀模型无法命中渠道自定义定价的问题。
2026-08-29 17:29:21 +08:00
wucm667 c31fe2ed9e fix(admin): preserve saved SMTP TLS mode in test endpoints
保留测试邮件端点已保存的 SMTP TLS 配置,区分省略与显式 false。
2026-08-29 17:28:00 +08:00
shaw 804679d996 fix(openai): preserve model scope on stream failover 2026-08-29 15:25:10 +08:00
shaw 571d1e1d9b fix(openai): isolate websocket semantic rate limits 2026-08-29 11:14:54 +08:00
shaw 3c22e78af0 fix(openai): preserve Spark quota reset semantics 2026-08-29 11:08:01 +08:00
shaw 5d9c7abed5 fix(openai): scope Spark quota 429 to model 2026-08-29 09:57:08 +08:00
Wesley Liddick ac18c588c8 Merge PR #6165
修复上游倍率探测导致账号列表整页刷新的问题
2026-08-29 09:12:05 +08:00
Wesley Liddick b83284071a Merge PR #5348
fix(subscription): 周/月窗口展示的重置时间与实际推进锚点对齐
2026-08-29 09:10:07 +08:00
Wesley Liddick fd13c72e95 Merge PR #6330
feat(zhipu): support team GLM Coding Plan usage query
2026-08-29 09:08:42 +08:00
Wesley Liddick 3c2e9a43b9 Merge PR #5860
fix(antigravity): preserve mixed built-in chat tools
2026-08-29 09:07:27 +08:00
Wesley Liddick ca319d09f2 Merge PR #6280
fix(update): strip hyphenated pre-release/custom suffix in version comparison
2026-08-29 09:06:10 +08:00
Wesley Liddick 46598dd491 Merge PR #6306
fix(openai): emit pre-output SSE keepalive on the /v1/responses passthrough path
2026-08-29 09:04:46 +08:00
Wesley Liddick 3045b3ade1 Merge PR #6316
fix(openai): honor Claude Code session id for messages sticky routing
2026-08-29 09:03:14 +08:00
Wesley Liddick 25dc5742ea Merge PR #6332
fix(monitor): 配额抓取在 singleflight 执行体内重查缓存,消除重复上游查询
2026-08-29 09:01:15 +08:00
Wesley Liddick 6506c0ea68 Merge PR #6343
fix(gateway): preserve Anthropic tool arguments
2026-08-29 09:00:56 +08:00
RayWangX 5688bcba91 fix(openai): honor Claude Code session for messages sticky routing 2026-08-29 08:31:15 +08:00
Jarl 1a61eb7151 test(usage): update compaction API contracts 2026-08-28 23:32:11 +08:00
Jarl 1cc6999ada feat(usage): track native compaction requests 2026-08-28 23:32:11 +08:00
风起 88cb79d8b3 fix(grok): honor client prompt_cache_key over X-Grok-Conv-Id for cache identity
grok-build CLI recap-style side-calls (turn-summary / title-refresh) send
a fresh per-call X-Grok-Conv-Id label but set body prompt_cache_key to
the parent session id, so a direct xAI connection shares the main turn's
server-side prompt cache prefix.

sub2api's explicitGrokCacheSeed preferred X-Grok-Conv-Id over the body
key, which fragmented the cache identity on every side-call and forced a
full-price replay of the whole conversation (~300K+ tokens each, ~2.3%
cache hit rate observed over 44 side-calls / 4.9M input tokens in 7
days — about a third of all grok-4.6 input).

Priority is now: CC session > generic session headers > body
prompt_cache_key > X-Grok-Conv-Id > explicitKey > previous_response_id.
IDE header-stability tests updated; new regression test locks in the
side-call/parent key-sharing contract.
2026-08-28 21:56:22 +08:00
Ryan ChouandClaude Opus 5 32ad1dcdce fix(subscription): 周/月窗口展示的重置时间与实际推进锚点对齐
订阅首窗初始化在开通日零点时,automaticWindowStartAt 会把实际推进锚点修正为
StartsAt,但展示层(calculateProgress / WeeklyResetTime / MonthlyResetTime)
仍按窗口起点+周期计算,导致仪表盘显示的重置时间早于实际重置时间(差值 =
StartsAt 的时分秒)。用户在显示的重置时间过后请求仍会收到 429
WEEKLY_LIMIT_EXCEEDED。

抽出 windowResetAnchor 供推进与展示共用,保证两者恒一致。非 legacy 锚点
(手动重置或已推进过的窗口)保持权威,不做修正。

日窗口不在本 PR 范围内:#5380 已把日额度改为按日历日对齐(DailyResetTime =
窗口起点所在日的次日 0 点),与周/月的期限对齐滚动窗口语义不同,不共用本修正。
附一个守护用例钉死这条边界。

Fixes #5348

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-28 21:46:44 +08:00