| #10788 | [Bug]: INFO logging of incoming requests to the LiteLLM proxy server can not be switched off. | Better | 192 / 78 | No new failures |
| #24677 | [Bug]: Incorrect TPM limiting for virtual keys | Better | 543 / 195 | No new failures |
| #25128 | [Bug]: MCP HTTP proxy client does not forward mcp-session-id to upstream stateful servers - tools never load | Better | 147 / 2,954 | No new failures |
| #25669 | [Bug]: Anthropic multi-turn tool call history destroyed when messages use Anthropic content format (tool_use/tool_result in content blocks) | Better | 542 / 128 | No new failures |
| #25857 | [Bug]: amazon.titan-embed-image-v1 — usage.prompt_tokens=0 for image inputs (EmbeddingResponse field, not cost calc) | Better | 107 / 6 | No new failures |
| #25889 | TypeError: '>' not supported between 'NoneType' and 'int' when Ollama returns rate limit error | Better | 171 / 2 | No new failures |
| #25947 | [Bug]: Vector stores from config.yaml are not added | Better | 146 / 941 | No new failures |
| #25985 | [Bug]: Ollama chat transformation fails with JSONDecodeError 'Unterminated string' when parsing tool call arguments | Better | 173 / 93 | No new failures |
| #26113 | [Bug]: remove_custom_field_from_tools() strips defer_loading — breaks Tool Search on Bedrock, wastes ~20k tokens per request | Better | 133 / 21 | No new failures |
| #26156 | [Bug]: Moonshot/Kimi K2.6 - reasoning_content is missing in assistant tool call message during multi-turn tool calling | Better | 201 / 77,108 | No new failures |
| #26939 | Title: Vertex AI Gemini streaming silently drops per-request timeout, falling back to litellm.request_timeout (6000s) | Better | 182 / 159 | Not run |
| #27363 | [Bug]: aembedding missing num_retries kwarg causes zero retries and no failover for embedding model groups | Better | 185 / 1 | No new failures |
| #36566 | [Bug]: litellm_content_filter evaluations missing from request logs and Guardrails Monitor | Worse | 259 / 590 | No new failures |
| #37877 | [Bug]: provider budget never counts embedding or rerank spend — RouterBudgetLimiting raises "custom_llm_provider is required" | Better | 197 / 147 | No new failures |
| #38547 | [Bug]: /v1/models never expands litellm_proxy/* wildcards — provider-endpoint discovery is gated on a static dict that excludes litellm_proxy | Better | 133 / 86 | No new failures |
| #38571 | [Bug]: Bedrock Realtime hides provider throttles as normal WebSocket closes and lacks disconnect cleanup | Better | 607 / 225 | No new failures |
| #40317 | [Bug]: openai dependency still pinned to <3.0.0 in latest release, blocks packages requiring openai>=3.0.0 | Worse | 29 / 3,006 | No new failures |
| #40477 | [Bug]: Gemini candidate with a finishReason but no content is dropped, so the reason never reaches the client | Better | 209 / 462 | No new failures |
| #40846 | [Bug]: Completed /v1/responses requests retain max_parallel_requests slots during deferred logging | Better | 222 / 199 | No new failures |
| #41202 | [Bug]: thinking-binding-controls-2026-08-01 beta header is stripped on /v1/messages, so thinking.block_binding 400s on Bedrock/Vertex | Better | 161 / 22 | No new failures |
| #41394 | [Bug]: All chatgpt/* models missing reasoning annotations present on their openai/ and azure/ twins | Better | 207 / 1,166 | No new failures |
| #41421 | [Bug]: LiteLLM Proxy 1.100.0 serializes image-edit uploads as strings for OpenAI GPT Image models | Better | 85 / 88 | No new failures |
| #41435 | [Bug]: Fix "vertex_ai/xai/grok-4.6" entry in "model_prices_and_context_window.json" — missing supports_prompt_caching | Better | 32 / 20 | No new failures |
| #41834 | [Bug]: Databricks service_tier is dropped in request translation, streaming, and pricing | Better | 335 / 56 | Not run |
| #41835 | [Bug]: Vertex Grok (vertex_ai/xai/grok-*) is rejected on /v1beta/models/{model}:generateContent while /v1/chat/completions on the same deployment works | Better | 180 / 2,768 | Not run |
| #42078 | [Feature]: Expose OpenTelemetry trace IDs on LiteLLM Proxy responses | Better | 380 / 1,192 | Not run |
| #42869 | [Bug]: streaming /v1/responses emits no reasoning item for signature-only thinking (Claude Fable 5.1 / Opus 5.5 default, Bedrock adaptive), so reasoning cannot be replayed | Better | 206 / 245 | No new failures |
| #43868 | [Bug]: Custom pricing silently bills ultrafast at standard rates despite published tier prices | Worse | 279 / 712 | No new failures |
| #44831 | [Bug]: MCP tools/call with an unprefixed tool name on /{server_name}/mcp resolves via the global tool-name mapping, not the server in the path (404 when cold, 403 when another server has a tool with the same name) | Better | 159 / 139 | Not run |
| #44921 | [Bug]: HiddenLayer JWT refresh blocks the event loop inside the async guardrail request | Worse | 98 / 142 | No new failures |