Skip to content

v0.1.85

Latest

Choose a tag to compare

@github-actions github-actions released this 02 Sep 12:07
· 6 commits to main since this release
f731e01

Changelog

Features

  • 0aec4dc feat(versioncheck): keep the install id in the database, with a derived fallback (#806)

Bug Fixes

  • 37a9a58 fix(admin): report effective resilience for dashboard-registered providers (#852)
  • 8e7d5a0 fix(gateway): make the chat fast path fire by default and honor the client's raw model (#849)
  • 65987f4 fix(llmclient): stop the retry backoff timer when the context is cancelled (#844)
  • 1cfa031 fix(modeldata): resolve metadata for models with a routing suffix (#815)
  • 7a8d6a6 fix(providers): never overlay a bare env key onto a renamed provider's explicit credentials (#850)
  • becf43f fix(routing): let the route selector decide session affinity (#821)
  • f731e01 fix(server): keep the gateway's rate-limit headers on fast-path chat responses (#854)
  • 26148dd fix(telemetry): default OpenTelemetry service.name to the product name (#814)

Performance

  • 44ef8c7 perf(gateway): fast-path non-streaming chat passthrough without dropping cache planning (#845)
  • d45d8f7 perf(providers): avoid JSON round trips in the prompt cache planner (#838)
  • f62e290 perf(providers): parallelize startup availability checks (#805)