All pull requests

Pull requests

• fix(github_copilot): surface copilot_usage token_details as cache token counts

  #48588·jeevan1098 opened Oct 27, 2026Contributor·4 of 6 tasks·

  ·
  1

  2

• fix(vertex-ai): drop the floor retirement date from gemini-3.1-flash-lite-image

  #48587·devin-ai-integration[bot] opened Oct 27, 2026Bot·

  ·
  2

• fix(router): raise RouterBudgetLimitError on crossed budget rejections

  #48586·yaseen2402 opened Oct 27, 2026·6 tasks done·

  ·
  1

  2

• feat(proxy): let operators turn provider pass-through routes off

  #48584·rice-as681 opened Oct 27, 2026·4 of 6 tasks·

  ·
  2

• fix(router): count concurrent successes atomically in usage-based routing

  #48583·charankotra348-glitch opened Oct 27, 2026·

  ·
  1

  2

• fix(bedrock): drop toolSpec.strict for Claude Fable 5

  #48582·TobiMayr opened Oct 27, 2026Contributor·4 of 6 tasks·

  ·
  2

• feat(providers): add Paxa as an OpenAI compatible speech provider

  #48581·yoyoismee opened Oct 27, 2026·3 of 5 tasks·

  ·
  2

• fix(langfuse): trace successful decisions calls with their token counts

  #48580·rodrigoazv opened Oct 27, 2026·4 of 6 tasks·

  ·
  1

  2

• fix(prometheus): chart spend log storage bytes in the all-metrics dashboard

  #48579·innovation-ways opened Oct 27, 2026·4 of 6 tasks·

  ·
  1

• fix(tests): use UTC datetimes in batch TPD reset test

  #48577·innovation-ways opened Oct 27, 2026·5 of 6 tasks·

  ·
  1

• fix(guardrails): prevent duplicate pre_call guardrail runs when metadata.user_id is present

  #48575·rajeshshanmugam-smf opened Oct 27, 2026·4 of 6 tasks·

  ·
  2

• fix(watsonx): enable use_in_pass_through support for watsonx passthrough endpoint.

  #48573·hasan4791 opened Oct 27, 2026Contributor·

  ·
  2

• fix(chatgpt): move system/developer input items into instructions

  #48566·sunhouy opened Oct 27, 2026·4 of 6 tasks·

  ·
  1

  2

• fix(gemini): pair streamed server-side tool calls with their responses

  #48565·FazeelUsmani opened Oct 27, 2026Contributor·3 of 6 tasks·

  ·
  1

  2

• fix: preserve vLLM choice fields and continuous streaming usage

  #48564·albertoperdomo2 opened Oct 27, 2026Contributor·3 of 6 tasks·

  ·
  3

• feat(providers): add BigModel support and model pricing

  #48563·AaronHowell opened Oct 27, 2026Contributor·

  ·
  2

• fix(modelscope): use async task polling for image generation

  #48558·yrk111222 opened Oct 27, 2026Contributor·6 tasks done·

  ·
  2

• feat(providers): add cheapaimodel as a JSON-configured OpenAI-compatible provider

  #48556·nicho810 opened Oct 27, 2026·4 of 6 tasks·

  ·
  2

• feat(search): add multi-engine SerpApi provider integration

  #48553·ValteruGowtham opened Oct 27, 2026·

  ·
  3

• fix(anthropic): report prompt tokens in message_start for bridged /v1/messages streams

  #48551·Gourav224 opened Oct 27, 2026·4 of 6 tasks·

  ·
  1

  2

• fix(gemini): keep the grounding citation for a span that starts at byte 0

  #48549·Kropiunig opened Oct 27, 2026Contributor·4 of 6 tasks·

  ·
  2

• fix(router): register deployments whose cost-map row has a non-chat mode

  #48548·sunhouy opened Oct 27, 2026·

  ·
  2

• feat(pricing): add Flint rates matching DeepInfra Qwen3.8-27B

  #48547·siilats opened Oct 27, 2026·3 of 5 tasks·

  ·
  2

• feat(watt): add watt cloud as a provider

  #48546·caoqiyong opened Oct 27, 2026·5 tasks done·

  ·

• docs: fix broken contributing links in README files

  #48542·yangjian7zhong opened Oct 27, 2026·

  ·
  2

  2