Skip to content

添加了图生文 混图路由 - #4

Merged
Calcium-Ion merged 3 commits into
QuantumNous:mainfrom
luxlzz6:main
Nov 14, 2023
Merged

添加了图生文 混图路由#4
Calcium-Ion merged 3 commits into
QuantumNous:mainfrom
luxlzz6:main

Conversation

@luxlzz6

@luxlzz6 luxlzz6 commented Nov 14, 2023

Copy link
Copy Markdown
Contributor

No description provided.

@Calcium-Ion
Calcium-Ion merged commit 8b2e4ec into QuantumNous:main Nov 14, 2023
@Calcium-Ion

Calcium-Ion commented Nov 14, 2023

Copy link
Copy Markdown
Member

感谢pr

Calcium-Ion pushed a commit that referenced this pull request Nov 19, 2023
@luxlzz6
Merge pull request #4 from Calcium-Ion/main
98326bc 
1
Calcium-Ion pushed a commit that referenced this pull request Jan 8, 2024
@luxlzz6
Merge pull request #4 from Calcium-Ion/main
a0fb3f7 
1
@jokenee jokenee mentioned this pull request Sep 18, 2024
Closed
@qiuwei qiuwei mentioned this pull request Aug 24, 2025
Closed
5 tasks
seefs001 pushed a commit that referenced this pull request Oct 10, 2025
@github-actions
Merge pull request #4 from QuantumNous/main
b516ca8 
Fork Sync: Update from parent repository
t0ng7u added a commit that referenced this pull request Oct 28, 2025
@t0ng7u
Merge pull request #4 from t0ng7u/feat/upstreamModelRatioSync
7099930 
feat: 上游倍率同步
sususu98 pushed a commit to sususu98/new-api that referenced this pull request Dec 1, 2025
@zsio
Feature/actions (QuantumNous#7)
5eb0fa1 
* 添加 GitHub Actions CI/CD 配置文件，包括 PR 构建检查和版本发布工作流，更新 README.md 以说明工作流和快速开始步骤。

* fix: initialize drizzle db lazily

* build: optimize docker image with standalone output

* build: speed up multi-arch docker build

* 优化 Dockerfile，移除镜像源前缀以简化构建过程

* chore: simplify release workflow and deployment docs

* fix: ensure release buildx uses container driver

* 优化 PR 检查工作流，使用单一架构以提高构建速度，并将镜像直接加载到 Docker daemon，简化后续测试步骤。

* 优化 PR 检查和发布工作流，增强镜像验证步骤，确保 Node.js 版本和应用文件存在性检查，改进版本标签匹配规则。

* Feature/actions (QuantumNous#3)

* 添加 GitHub Actions CI/CD 配置文件，包括 PR 构建检查和版本发布工作流，更新 README.md 以说明工作流和快速开始步骤。

* fix: initialize drizzle db lazily

* build: optimize docker image with standalone output

* build: speed up multi-arch docker build

* 优化 Dockerfile，移除镜像源前缀以简化构建过程

* chore: simplify release workflow and deployment docs

* fix: ensure release buildx uses container driver

* 优化 PR 检查工作流，使用单一架构以提高构建速度，并将镜像直接加载到 Docker daemon，简化后续测试步骤。

* 优化 PR 检查和发布工作流，增强镜像验证步骤，确保 Node.js 版本和应用文件存在性检查，改进版本标签匹配规则。

* 删除 GitHub Actions 配置的 README.md 文件，移除不再需要的文档内容。

* 删除 action 中的 readme (QuantumNous#4)

* 添加 GitHub Actions CI/CD 配置文件，包括 PR 构建检查和版本发布工作流，更新 README.md 以说明工作流和快速开始步骤。

* fix: initialize drizzle db lazily

* build: optimize docker image with standalone output

* build: speed up multi-arch docker build

* 优化 Dockerfile，移除镜像源前缀以简化构建过程

* chore: simplify release workflow and deployment docs

* fix: ensure release buildx uses container driver

* 优化 PR 检查工作流，使用单一架构以提高构建速度，并将镜像直接加载到 Docker daemon，简化后续测试步骤。

* 优化 PR 检查和发布工作流，增强镜像验证步骤，确保 Node.js 版本和应用文件存在性检查，改进版本标签匹配规则。

* 删除 GitHub Actions 配置的 README.md 文件，移除不再需要的文档内容。

* 优化 PR 检查工作流，移除不必要的 Docker 镜像结构测试步骤，以简化构建过程。

* 删除 GitHub Actions 配置中的 README.md 文件，移除不再需要的文档内容。

* 更新 PR 检查工作流，仅在向 main 分支提交 PR 时触发，简化触发条件。
sususu98 pushed a commit to sususu98/new-api that referenced this pull request Dec 1, 2025
@ding113 @claude
fix: validate canLoginWebUi permission in middleware (QuantumNous#4)
1866fac 
修复 key 不允许登陆 web 面板的设置无效的问题

**问题描述**：
- 登录接口正确检查了 canLoginWebUi 权限
- 但中间件只检查 cookie 是否存在，没有重新验证 key 的权限
- 导致用户登录后，管理员修改 canLoginWebUi=false 时，用户仍可访问 dashboard

**修复方案**：
- 在 middleware 中调用 validateKey() 验证完整权限
- 包括 canLoginWebUi、isEnabled、expiresAt 等检查
- 如果验证失败，清除 cookie 并重定向到登录页

Fixes QuantumNous#3

Co-authored-by: claude[bot] <209825114+claude[bot]@users.noreply.github.com>
Co-authored-by: Ding <ding113@users.noreply.github.com>
DeJeune pushed a commit to DeJeune/new-api that referenced this pull request Jan 27, 2026
@zhaolion @claude
fix: exclude rerank models from embedding detection in channel test (Q…
f4471c8 
…uantumNous#4)

* feat: enhance channel testing with extended model type detection

- Exclude image edit models from testing (not supported)
- Add endpoint type tracking for embeddings and rerank models
- Add image generation detection for qwen and gpt image models
- Add response endpoint detection for gpt-5.2 models

Co-Authored-By: Claude <noreply@anthropic.com>

* feat: add kwai-kolors image generation model detection

Add endpoint detection for kwai-kolors image generation models
to use /v1/images/generations endpoint during channel testing.

Co-Authored-By: Claude <noreply@anthropic.com>

* fix: exclude rerank models from embedding detection in channel test

Rerank models like BAAI/bge-reranker-v2-m3 contain "bge-" in their name
and were incorrectly matched as embedding models. This fix ensures
rerank models are properly routed to the /v1/rerank endpoint.

Co-Authored-By: Claude <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
zonde306 pushed a commit to zonde306/new-api that referenced this pull request Feb 12, 2026
zonde306
refactor: plans/priority-0-rate-limit-redis.md task QuantumNous#4
9ee1925
zonde306 pushed a commit to zonde306/new-api that referenced this pull request Feb 12, 2026
zonde306
refactor: priority-1-db-write-batch.md QuantumNous#4
921bfbf
zonde306 pushed a commit to zonde306/new-api that referenced this pull request Feb 13, 2026
zonde306
refactor: priority-2-streaming-sse.md QuantumNous#4
2357369
zonde306 pushed a commit to zonde306/new-api that referenced this pull request Feb 14, 2026
zonde306
refactor: priority-3-routing-cache.md QuantumNous#4
9884511
cuijinyu referenced this pull request in cuijinyu/new-api Mar 11, 2026
@cuijinyu
Merge pull request #4 from gallon666/codex/asynchronously-write-llm-l…
a8dbc37 
…ogs-to-aws-s3

feat: asynchronously write LLM raw request/response logs to AWS S3
smysle added a commit to smysle/new-api that referenced this pull request Mar 29, 2026
@smysle
fix: 修复邀请返利的幂等性、原子性和校验问题
2d88845 
🔴 Critical QuantumNous#1: 管理员补单重复返利
- ManualCompleteTopUp 返回 (completed, creditedQuota, topUpId, userId, err)
- 已成功订单返回 completed=false，不再触发返利
- 正确处理 Creem 订单的额度计算（Amount 即额度）

🔴 Critical QuantumNous#2: Webhook 重试导致重复返利
- TopUp 模型新增 InviterRewardSent 字段（GORM AutoMigrate）
- ProcessInviterReward 增加 topUpId 参数，通过原子 UPDATE 实现幂等
- 所有调用方传入 topUpId（兑换码场景传 0 跳过检查）

🟠 Major QuantumNous#3: 改 Type 时不检查当前 Value 兼容性
- InviterRewardType 切换到 percentage 时，校验当前 value ≤ 100

🟠 Major QuantumNous#4: 非原子更新 + 吞错误
- 移除双重 quota 更新（IncreaseUserQuota），只更新 aff_quota/aff_history
- aff_* 更新失败时返回 error 而非吞掉

Nitpick QuantumNous#1: InputNumber 清空时 String(null) → String(value ?? '')
Nitpick QuantumNous#2: API 响应检查 res.some(r => !r || !r.success)
quqi1599 referenced this pull request in quqi1599/new-api Apr 6, 2026
@rebecca554owen
Merge pull request #4 from QuantumNous/main
05e0803 
合并上游更新
stratosphere-ai pushed a commit to stratosphere-ai/new-api that referenced this pull request Apr 12, 2026
@claude
fix: resolve 10 audit issues (4 critical + 6 important)
7d1cbc3 
Critical fixes:
- QuantumNous#1+QuantumNous#10: Wire hourly/daily cap checking into channel selection
  via ChannelShouldSkip hook + listing cache for hot-path lookups.
  IncrementHourlyDailyCaps now called after successful trade.
- QuantumNous#2: Capture RecordTrade error in relay handler and log it
- QuantumNous#3+QuantumNous#4+QuantumNous#6: Wrap RecordTrade in GORM DB.Transaction for atomicity
  (trade record + seller credit + token count in single tx, rollback
  on any failure)

Important fixes:
- QuantumNous#5: Add seller self-service withdrawal request endpoint
  POST /seller/withdrawal/request with CriticalRateLimit
  + withdraw button on earnings page
- QuantumNous#7: Admin suspend now collects and reports listing disable errors
  instead of silently ignoring them
- QuantumNous#8: Anthropic key validation changed from POST /v1/messages to
  GET /v1/models (lightweight, read-only, no token consumption)
- QuantumNous#9: CriticalRateLimit added to seller register, create listing,
  create API token, and withdrawal request endpoints

https://claude.ai/code/session_01VUyzgwX29o57mjURxJvYiN
x22x22 pushed a commit to x22x22/new-api that referenced this pull request Apr 24, 2026
@Calcium-Ion
Merge pull request QuantumNous#4 from luxlzz6/main
e25da16 
添加了图生文 混图路由
x22x22 pushed a commit to x22x22/new-api that referenced this pull request Apr 24, 2026
@luxlzz6
Merge pull request QuantumNous#4 from Calcium-Ion/main
b3aafb4 
1
@aiken aiken mentioned this pull request May 14, 2026
Closed
k272314374 pushed a commit to k272314374/hmrouter that referenced this pull request May 16, 2026
@kibbin90 @claude
页面评审修补 — 5 处文档问题收口
828e2d4 
另一窗口页面评审发现 5 处，逐条修复：

QuantumNous#1 TASKS M1 任务数 74→104（实质 bug）
   - 燃尽快报表 / 章节标题 / 关键指标 / 变更日志 四处 74 → 104
   - 74 是 9 通道拆分前的旧数；9 通道实际 = DB13+BA10+BB14+BC10+BD13
     +FA11+FB13+FC11+OP4+基础设施5 = 104（工作量汇总总计 117 = 9+104+4 反证）
   - STATE "共 100 任务" → 燃尽 130（M0 12+Spec 5+Phase0 9+M1 104）
   - 新增 CHK-15：燃尽 M1 数 == M1 章节实际任务行数（防再次失配）

QuantumNous#2 PRD §3 "员工账户偏好"行勾选不一致
   - 5 管理角色裸 ✅ → 统一 "✅自己"（管理员只能改自己偏好）

QuantumNous#3 PAGES §1 命名图例漏 AA 域码
   - 补 "AA=Admin-action Audit 元审计"

QuantumNous#4 STATE 未记录 5/16 改动
   - 头部日期 5/15→5/16；新增 "S-003 续（2026-05-16）" session 记录
     （coding-principles / 四层对齐修补 / 仓库重组）
   - 顺修：第三处编辑误删的 S-002 标题已还原

QuantumNous#5 /admin→/enterprise 路由迁移跨文档残留
   - ARCHITECTURE §7 features 树：admin/{announcements,skills,courses}
     → enterprise/{announcements,skills,courses}
   - agent-collaboration.md §1 表 + §8 CODEOWNERS 同步
     （health/dept-model-acl/system-settings 仍归 admin/ 系统域）
   - CODEOWNERS 加 enterprise/{announcements,skills,courses} → frontend-c 具体行

一致性 15/15 ✅

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
jiutubaba pushed a commit to jiutubaba/fx-api that referenced this pull request May 17, 2026
@Wei-Shaw
Merge pull request QuantumNous#4 from NepetaLemon/refactor/backend-wi…
0e7b8aa 
…re-provider-sets

refactor(backend): 拆分 Wire ProviderSet
jiutubaba pushed a commit to jiutubaba/fx-api that referenced this pull request May 17, 2026
@touwaeriol
fix(payment): critical audit fixes for security, idempotency and corr…
c738cfe 
…ectness

Backend fixes:
- QuantumNous#1: doSub subscription idempotency via audit log check
- QuantumNous#2: markFailed only when status=RECHARGING (prevents overwriting COMPLETED)
- QuantumNous#3: ExpireTimedOutOrders checks upstream payment before expiring
- QuantumNous#4: Public verify endpoint for payment result page (no auth required)
- QuantumNous#5: EasyPay QueryOrder returns amount, confirmPayment handles zero amount
- QuantumNous#6: WxPay notifyUrl priority: request-first, config-fallback
- QuantumNous#7: EasyPay remove double URL decode in VerifyNotification
- QuantumNous#8: checkPaid/cancelUpstreamPayment use order's provider instance
- QuantumNous#9: Amount NaN/Inf/negative validation in order creation and refund
- QuantumNous#10: Refund amount comparison uses tolerance instead of float64 ==
- QuantumNous#11: Skip balance deduction on retry when previous rollback failed
- QuantumNous#12: checkPaid logs fulfillment errors instead of silently ignoring
- QuantumNous#13: WxPay certSerial added to required config fields

Frontend fixes:
- Payment result page no longer requires authentication
- Public verify API fallback for expired sessions
yujipeng pushed a commit to yujipeng/new-api that referenced this pull request May 21, 2026
jipeng.yu @multica-agent
docs(operations): close 12 pending items with evidence-backed wording
f09f6a4 
Round 2 sync per architect's evidence closure (TES-69 comment 87c3958d):

- Rewrite 99-pending-items.md: all 12 grey-zone items closed-2026-05-21
  with evidence chains + handling outcome; appendix lists 5 known
  improvements (audio/task gate parity, svip default, three frontend
  guards, RPM/TPM tooltip, quota_data cleanup) tracked by PJM separately
- 03-billing.md #1: add 3.1.4 explaining text-path two gates
  (SelfUseMode + AcceptUnsetRatioModel) vs audio/task path no-gate
  asymmetry for unregistered models
- 03-billing.md QuantumNous#2 (in 02): rewrite TopupGroupRatio as
  topup-side discount/markup multiplier, not billing-side; document
  formula, four payment paths, fallback 1.0 behavior
- 03-billing.md QuantumNous#9: rewrite OtherRatios as active dynamic billing
  multiplier for task/image paths (seconds/size/n), list adapters,
  clarify text-path no-op semantics; remove "reserved field" wording
- 01/02 QuantumNous#3 QuantumNous#4 QuantumNous#5 QuantumNous#6 QuantumNous#7: operational hints planted (group ref check,
  FixAbility low-peak single-instance, double-write reminder w/ note
  that automatic linkage check is not yet implemented, svip default
  workaround, ip log default-false with compliance rationale)
- 05-faq QuantumNous#8 errata: actual error message is
  MsgDistributorTokenModelForbidden ("该 token 不允许使用模型 X"),
  not "token model limit is empty"; FAQ scene 4 rewritten as full
  troubleshooting chain
- 05-faq QuantumNous#10 QuantumNous#11 QuantumNous#12: planted hints for usingGroup-only logs.group
  (open other.user_group / other.token_group after schema extension),
  RPM/TPM as 60s snapshot, quota_data cleanup via DBA
- README, 5.2 summary table, 5.3 troubleshooting matrix, and inline
  anchors updated to match new 99-pending-items.md headings

Co-authored-by: multica-agent <github@multica.ai>
@giovannimnz giovannimnz mentioned this pull request Jun 2, 2026
Closed
7 tasks
zer0Black added a commit to Agents-Zone/sili-smart-api that referenced this pull request Aug 1, 2026
@zer0Black @claude
refactor(conversation): 批次 Minor QuantumNous#4 空名工具调用跳过
ca9db9d 
Co-Authored-By: Claude <noreply@anthropic.com>
leikaiwei added a commit to leikaiwei/new-api that referenced this pull request Aug 7, 2026
@leikaiwei
docs: README 增加补丁 QuantumNous#4 说明
4f4e594 
记录两套缓存语义的差异、闸门为何恒不触发、影响面精确到「上游报
cached_tokens 但不报 cache_write_tokens」这一格，以及计费路径不受影响
与反向转换不会双减的核对结论。

定位判据（不同上游路径的 cache_read/text_tokens 比值：经 new-api 恒定
90.2%、真 Anthropic 上游可达 945.9%）标注为取自 LiteLLM 侧日志统计而非
本仓库实测；未直接抓取 new-api 出站 input_tokens 原值这一限制也一并记下，
正确性依据是语义分析加 golden 与单测。

另记为何配置层绕不开：param_override 的 11 个调用点全部作用于出站请求体，
无响应侧覆盖钩子；换 LiteLLM 入口为 OpenAI 风格会让补丁 QuantumNous#2 的
client_thinking_type 与补丁 QuantumNous#3 一并失效。
Luckylos pushed a commit to Luckylos/new-api that referenced this pull request Aug 14, 2026
Luckylos Dev @Luckylos
refactor(relay): pay down failover tech debt — multi-key aware exclud…
77c3bac 
…e, pure exclude-driven auto group, adaptive cap fixes

- Remove dead `retry` param from GetRandomSatisfiedChannel/GetChannel (debt QuantumNous#1)
- Rewrite auto-group selection to be purely exclude-driven; drop priorityRetry
  index, retry-counter manipulation, resetNextTry, and dead AutoGroupRetryIndex
  writes. Distinguish discovery miss (always skip group) from failover
  exhaustion (advance only when cross-group retry enabled) (debt QuantumNous#2)
- Multi-key aware exclude: a channel is only excluded after all its enabled
  keys have been tried, preserving per-request key rotation (debt QuantumNous#3)
- Size retry budget by total enabled keys (CountEnabledKeys) so every key of a
  multi-key channel fits within the adaptive cap (debt QuantumNous#3)
- Fix shouldRetry to use retryCap instead of common.RetryTimes so the adaptive
  cap is not silently clamped (debt QuantumNous#4)
- Task relay loop now carries exclude set with the same multi-key logic
- Tests: multi-key CountEnabledKeys + key-weighted budget
@lywaterman lywaterman mentioned this pull request Aug 18, 2026
Merged
9 tasks
leikaiwei added a commit to leikaiwei/new-api that referenced this pull request Aug 19, 2026
@leikaiwei
docs: README 增加补丁 QuantumNous#4 说明
39e5cc6 
记录两套缓存语义的差异、闸门为何恒不触发、影响面精确到「上游报
cached_tokens 但不报 cache_write_tokens」这一格，以及计费路径不受影响
与反向转换不会双减的核对结论。

定位判据（不同上游路径的 cache_read/text_tokens 比值：经 new-api 恒定
90.2%、真 Anthropic 上游可达 945.9%）标注为取自 LiteLLM 侧日志统计而非
本仓库实测；未直接抓取 new-api 出站 input_tokens 原值这一限制也一并记下，
正确性依据是语义分析加 golden 与单测。

另记为何配置层绕不开：param_override 的 11 个调用点全部作用于出站请求体，
无响应侧覆盖钩子；换 LiteLLM 入口为 OpenAI 风格会让补丁 QuantumNous#2 的
client_thinking_type 与补丁 QuantumNous#3 一并失效。
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@luxlzz6 @Calcium-Ion