All pull requests

Pull requests

• fix(ui): show zero-spend days in the usage page daily spend chart

  #43025·amitdey1987 opened Sep 24, 2026·

  ·
  3

• fix(presidio): mask streamed /v1/messages output when the first upstream read is a keepalive, a data-less ping, or a split utf8 character

  #43023·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  9

  

• fix(s3_v2): drop terminal upload failures, budget transient retries per flush and bound the queue at enqueue

  #43022·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  25

  

• docs(github): require UI before/after screenshots and intentional UX change note in PR template

  #43021·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  8

  

• fix(proxy): traceparent/baggage fallback must not override caller metadata

  #43020·fandujar opened Sep 24, 2026·

  ·
  19

• fix(proxy): enforce key max_budget from budget_id table

  #43018·vineoy opened Sep 24, 2026·

  ·
  1

  8

• fix(router): serve Responses turns from a sibling when the encrypted content origin has no boundary peer

  #43015·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  1

  11

• fix(ui): group cost optimization cache leakage by model group

  #43008·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  7

  

• chore(backport): carry the deferred native stream dispatch fix to stable/1.101.x

  #43007·keppy opened Sep 24, 2026·

  ·
  2

• fix(logging): dedupe failure callbacks on streaming requests

  #43004·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  1

  7

• fix(bedrock): strip Anthropic native extensions Bedrock Invoke rejects from Messages bodies

  #43001·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  3

  

• feat: expose litellm.__version__

  #42999·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  5

  

• fix(proxy): delete large teams without per-member transaction fan-out

  #42998·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  7

• feat(rust): add a native Cache facade behind a PYTHON_ONLY catalog rule

  #42997·yujonglee-berri opened Sep 24, 2026Contributor·

  ·
  10

• feat(proxy): server-side Team Usage export beyond the top-N key cap

  #42996·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  21

  

• fix(lakera_ai): keep flagged verdict as blocking floor with category_thresholds

  #42993·Varshith-Kali opened Sep 24, 2026·

  ·
  1

  5

• refactor(redis): stop proxy code driving RedisCache's raw redis-py client

  #42991·yujonglee-berri opened Sep 24, 2026Contributor·

  ·
  11

• fix(gemini): translate Anthropic messages for countTokens and fall back locally on provider errors

  #42990·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  3

  

• feat(terraform): add display_name to litellm_model resource and model data sources

  #42987·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  16

  

• feat(caching): give RedisCache a pub/sub and lease surface and drop raw client reaches

  run-ci
  #42984·devin-ai-integration[bot] opened Sep 24, 2026Bot·

  ·
  9

  

• fix(mcp): surface upstream OAuth error on 200 token response without access_token

  #42983·soyasticks opened Sep 24, 2026·

  ·
  1

  11

• fix(proxy): enforce Responses metadata tags in native budgets

  #42979·luisgallego-aily opened Sep 24, 2026Contributor·

  ·
  2

• fix(responses): avoid duplicate spend logs in async chat bridge

  #42977·sorgfresser opened Sep 24, 2026·

  ·
  4

• fix(anthropic): omit top_p from supported params for Claude models

  #42975·IsmaelMartinez opened Sep 24, 2026·

  ·
  1

  6

• feat(staan): add Staan.ai search provider support

  #42973·TobiasGoerke opened Sep 24, 2026·

  ·
  8