Version 0.1.5

Features

• Log Management: Batch log cleaning with optimized performance (#90, #89, #76)
• Database Optimization: Asynchronous log index creation & vacuum logs (#85, #84)
• Dashboard Enhancements: Timestamp query support & token usage statistics (#83, #82, #69)
• Security: IP group detection and blocking (#79)
• Model Configurations: Group model configuration support (#70)
• API Improvements:
  • VertexAI Claude cache usage tracking (#68)
  • Claude extended tool type support (#63)
  • Basic AIProxy HTTP handler (#64)
  • Channel parameter support (#77)
• Monitoring:
  • TTFB (Time to First Byte) recording (#73)
  • RPM/TPM metric optimizations (#81)

Fixes

• Data Cleanup:
  • Concurrent cleanup fix (#93)
  • SQL query optimizations (#92, #86)
  • Decimal summation for used amounts (#88)
• API Handling:
  • VertexAI Claude stream mode input usage (#65)
  • Gemini image detection (#52)
  • Azure API version defaults (#42)
• Logging:
  • RequestID exclusion from log indexes (#91)
  • Time truncation adjustments (#87)
• Billing:
  • PDF price parsing (#37)
  • Token expression fixes (#27)

Performance

• Log index optimizations (#91, #74)
• Database query optimizations (#71)
• Cleanup process performance improvements (#89, #86)

Chores

• Documentation:
  • Updated usage documentation (#80)
  • Swagger API docs implementation (#50, #75)
  • Listen address customization (#48)
• Dependencies:
  • JWT library bump to v5.2.2 (#66)
  • AWS credentials dependency update (#72)
• Code Quality:
  • Removed unused packages (#36)
  • SQL query optimizations (#71)
  • Error handling improvements (#23, #28)

Infrastructure

• Memory limits and header constraints (#43)
• Token billing cache support (#33)
• Embed tokenization optimization (#34)

Version 0.1.5-beta.6 (Prerelease)

v0.1.5-beta.6

fix: tokens omit empty

Version 0.1.5-beta.5 (Prerelease)

v0.1.5-beta.5

readme: update usage (#80)

Version 0.1.5-beta.4 (Prerelease)

Feat:

• claude more tool type support #63
• basic aiproxy http handler #64
• vertexai claude cache usage #68
• dashboard token usage #69
• group model config #70
• optimize log sql #71

Fix:

• add group token swag query docs #59
• rpm detail group can be empty #60
• dashboard rpm detail from cache #61
• vertex ai claude stream mode dosnot resp input usage #65

Version 0.1.5-beta.3 (Prerelease)

Feat:

• full swat params #58
• channel group sets support #57

Fix:

• price and usage typed #55
• check gemini has image #52

Version 0.1.5-beta.2 (Prerelease)

Feat:

• swagger docs #50
• docs: update listen readme #49

Version 0.1.5-beta.1 (Prerelease)

Feat:

• custom listen addr #48

Fix:

• xunfei returns the input modelname when no case matches for obtaining the model domain #46

Version 0.1.4

Feat:

• open router reasioning to reasoning_content #44
• rpm mem limit and limit header #43
• embed tiktoken #34
• cache token billing support #33
• grok api and error handler #32
• gemini image #31
• log query result onl #30
• baiduv2 rerank #26

Fix:

• exclude tests model #41
• update azure default api version #42
• dependabot no need push image #39
• parse pdf price #37
• clean modeltype and clean no used package #36
• need write status code when context done #28
• baidu and vertexai token expr #27
• claude handle balance error #23

Version 0.1.3

Fix:

• ali rerank model #21

Version 0.1.2

Fix:

• ollama tool call #19
• ali stt realtime #20