●Change8

v1.80.10-nightly

📦 litellmView on GitHub →
✨ 26 features🐛 28 fixes🔧 31 symbols

Summary

Litellm v1.80.10 introduces Azure GPT‑5.2 model support, new security guardrail evidence, JWT team‑id selection, OTEL latency metrics, UI enhancements, and a large set of bug fixes and documentation updates.

✨ New Features

  • Add Azure GPT‑5.2 models support
  • Add opt‑in evidence results for Pillar Security guardrail during monitoring
  • Add support for expires after parameter in Files endpoint
  • Add support for forwarding client headers in /rerank endpoint
  • Add v0 support for target storage and documentation
  • New badge for Agent Usage in UI
  • JWT authentication now allows selecting team_id from request header
  • Add latency metrics (TTFT, TPOT, Total Generation Time) to OTEL payload
  • Add benchmark_proxy_vs_provider.py script
  • UI now shows version on top left near logo
  • Support reasoning_effort='xhigh' for GPT‑5.2 models
  • UI playground allows custom model name as first option
  • Support using MCPs on /chat/completions
  • Re‑organize UI left navigation and expose agents on root
  • Add Usage Entity labels
  • Add EU Claude Opus 4.5 model for Bedrock
  • Add all proxy models to default user settings
  • Add agent cost tracking on UI
  • A2A Gateway now allows adding Azure Foundry agents via UI
  • Show progress and pause on hover for notifications
  • Add extra_headers and allowed_tools to UpdateMCPServerRequest
  • Add storage_backend and storage_url columns to Prisma schema
  • Add import image to A2A documentation
  • Add agent_id field to GCS PubSub spend_logs_payload.json test expectation
  • Add Regex guardrails update
  • Add Cursor integration documentation

🐛 Bug Fixes

  • Add semver prerelease suffix to Helm chart versions
  • Add speechConfig to GenerationConfig for Gemini TTS
  • Add 'exception_status' field to Prometheus logger
  • Handle string content in is_cached_message cache check
  • Fix token array input decoding for embeddings
  • Add minimum request threshold for error‑rate cooldown in router
  • Fix Milvus client documentation
  • Make create_litellm_branch tool more robust
  • Add health endpoint tests with database and Redis support
  • Use Docker executor for CI
  • Remove dependency on database and Redis from health test
  • Replace time.perf_counter() with time.time() in router
  • Fix x‑litellm‑key‑spend header update
  • Fix Bedrock header forwarding with custom API
  • Fix UI playground custom model name option issue
  • Fix UI Agent Usage page minor issues
  • Fix MCP tool name prefix
  • Fix CI/CD mypy, check_code_and_doc_quality, and MCP testing
  • Fix failing proxy unit test and Langfuse trace_id test
  • Fix failing proxy and core integration tests
  • Fix JWT authentication tests failing with KeyError: 'headers'
  • Fix UI request and response logs display
  • Fix Bedrock tool calling test failures with non‑serializable objects and internal parameters
  • Fix filter internal params in fallback code
  • Fix A2A Gateway Azure Foundry agents functionality
  • Fix schema.prisma files to include storage_backend and storage_url columns
  • Fix GCS PubSub spend_logs_payload.json test expectation for agent_id
  • Fix links in documentation

Affected Symbols

GenerationConfigPrometheusLoggeris_cached_messageEmbeddingHandlerRoutercreate_litellm_branchHealthEndpointDockerExecutorFilesEndpointHeaderHandlerBedrockClientRerankEndpointTargetStorageAgentUsageBadgeJWTAuthOTELPayloadBenchmarkProxyVsProviderUIVersionDisplayOpenAIModelConfigUIPlaygroundMCPHandlerUILeftNavUsageEntityBedrockModelUserSettingsCostTrackingUIAzureFoundryAgentUIDocusaurusThemeMermaidCIConfigRegexGuardrailCursorIntegration