Issues

• [Bug]: Fallback budget check does not cover team-member budget, so a zero-cost model with a paid fallback still spends past the cap

  llm translation
  BerriAI/litellm#46874 · Filter by author AlvaroGG0 opened on Oct 16, 2026

• [Bug]: Cannot edit a model with configured access windows

  bug
  BerriAI/litellm#46872 · Filter by author IAluI opened on Oct 16, 2026
  1

• [Bug]: Azure Responses API sends the gpt5_series/ routing prefix to Azure as the deployment name

  llm translation
  BerriAI/litellm#46866 · Filter by author ClemDNL opened on Oct 16, 2026
  2

• [Bug]: aiohttp per-host pool exhaustion is silent, requests just time out with no diagnostic

  llm translation
  BerriAI/litellm#46864 · Filter by author Maluslock opened on Oct 16, 2026

• [Bug]: free tier SSO lets a 6th user in, then locks out every SSO user including admins

  bug
  BerriAI/litellm#46853 · Filter by author LuksFlukss opened on Oct 16, 2026

• [Feature]: Support Bedrock API keys (bearer token) for Bedrock rerank

  enhancementllm translation
  BerriAI/litellm#46850 · Filter by author onceupon opened on Oct 16, 2026

• [Bug]: case-sensitive header lookup in _get_extra_header_tags drops configured spend tags

  bug
  BerriAI/litellm#46847 · Filter by author Dotify71 opened on Oct 16, 2026
  1

• [Bug]: aimage_edit() turns a missing image into [None] instead of [], crashing Black Forest Labs image edit with an uncaught ValueError

  llm translation
  BerriAI/litellm#46835 · Filter by author ryuto-takemoto opened on Oct 16, 2026
  2

• [Feature]: "Key Usage" view on the Usage page (filter by virtual key, with token metrics)

  enhancementllm translation
  BerriAI/litellm#46804 · Filter by author gmourahub opened on Oct 16, 2026
  1

• [Bug]: Gemini embeddings read text that starts with data:, gs:// or files/ as media, with no way to send it as text

  llm translation
  BerriAI/litellm#46719 · Filter by author georgeh0 opened on Oct 15, 2026

• [Bug]: Azure AI Foundry Anthropic endpoints 400 when azure_ad_token leaks into request body

  llm translation
  BerriAI/litellm#46701 · Filter by author emerzon opened on Oct 15, 2026

• [Feature] Per-MCP-server opt-out for the mandatory tool-name prefix (preserve exact tool names)

  llm translation
  BerriAI/litellm#46692 · Filter by author KK291860 opened on Oct 15, 2026

• [Bug]: Gemini/Vertex Realtime silently drops client tools and instructions with eager setup

  llm translation
  BerriAI/litellm#46685 · Filter by author emerzon opened on Oct 15, 2026

• [Bug]: Responses translation drops Anthropic tool_reference blocks from function call outputs

  llm translation
  BerriAI/litellm#46681 · Filter by author grainnemcknight opened on Oct 15, 2026

• [Bug]: Headroom compression and prompt-caching savings recorded as $0 for openrouter/* models since cost-map rows were removed

  BerriAI/litellm#46677 · Filter by author alike03 opened on Oct 15, 2026
  1

• [Bug]: RouterRateLimitError in litellm/router.py

  bugllm translation
  BerriAI/litellm#46669 · Filter by author lucas-profoundvision opened on Oct 15, 2026
  1

• [Feature]: Native support for OpenRouter's In-Region Routing (us.openrouter.ai / eu.openrouter.ai)

  llm translation
  BerriAI/litellm#46665 · Filter by author NinjaEde opened on Oct 15, 2026

• [Bug]: BYOK/static-auth MCP tools/list rejects per-request x-mcp-{alias}-authorization credential

  bug
  BerriAI/litellm#46664 · Filter by author morganrowse opened on Oct 15, 2026
  2

• [Bug]: v1.105.0 GZipBufferedResponseMiddleware withholds response headers of streamed (SSE) responses until the first chunk; MCP tools/call streams start only at the 15s ping

  llm translation
  BerriAI/litellm#46659 · Filter by author umm0n opened on Oct 15, 2026
  2

• [Bug]: MCP: cancelling a request handler during an in-flight stdio tools/call (session DELETE or notifications/cancelled) pins the proxy event loop at 100% CPU and leaks the server process (v1.104.0 through v1.105.1)

  llm translation
  BerriAI/litellm#46657 · Filter by author umm0n opened on Oct 15, 2026
  2

• [Bug]: Deployment custom pricing is billed correctly but callback and spend-log model-map rates use backend prices

  llm translation
  BerriAI/litellm#46651 · Filter by author sudhanshugupta-p72 opened on Oct 15, 2026
  1

• complexity_router: OpenCode subagent injections count as fresh human asks

  llm translation
  BerriAI/litellm#46646 · Filter by author Davide95 opened on Oct 15, 2026
  1

• MCP gateway DCR rejects Gemini app connectors: redirect_uris must be a list of 1 to 3 URIs

  llm translation
  BerriAI/litellm#46640 · Filter by author madsbergstroem opened on Oct 15, 2026

• [Bug]: Enterprise-only error when adding tags during virtual key creation in OSS

  bug
  BerriAI/litellm#46634 · Filter by author yash30389 opened on Oct 15, 2026

• [Bug]: Proxy 429 responses on /v1/messages drop the retry-after header that /v1/chat/completions preserves

  llm translation
  BerriAI/litellm#46630 · Filter by author xinliu0727-sudo opened on Oct 15, 2026
  1