BerriAI/litellm v1.67.0-stable

on GitHub

What's Changed

• build(model_prices_and_context_window.json): add gpt-4.1 pricing by @krrishdholakia in #9990
• [Fixes/QA] For gpt-4.1 costs by @ishaan-jaff in #9991
• Fix cost for Phi-4-multimodal output tokens by @emerzon in #9880
• chore(docs): update ordering of logging & observability docs by @marcklingen in #9994
• Updated cohere v2 passthrough by @krrishdholakia in #9997
• [Feat] Add support for cache_control_injection_points for Anthropic API, Bedrock API by @ishaan-jaff in #9996
• [UI] Allow setting prompt cache_control_injection_points by @ishaan-jaff in #10000
• Fix azure tenant id check from env var + response_format check on api_version 2025+ by @krrishdholakia in #9993
• Add /vllm and /mistral passthrough endpoints by @krrishdholakia in #10002
• CI/CD fix mock tests by @ishaan-jaff in #10003
• Setting litellm.modify_params via environment variables by @Eoous in #9964
• Support checking provider /models endpoints on proxy /v1/models endpoint by @krrishdholakia in #9958
• Update AWS bedrock regions by @Schnitzel in #9430
• Fix case where only system messages are passed to Gemini by @NolanTrem in #9992
• Revert "Fix case where only system messages are passed to Gemini" by @krrishdholakia in #10027
• chore(docs): Update logging.md by @mrlorentx in #10006
• build(deps): bump @babel/runtime from 7.23.9 to 7.27.0 in /ui/litellm-dashboard by @dependabot in #10001
• Fix typo: Entrata -> Entra in code by @msabramo in #9922
• Retain schema field ordering for google gemini and vertex by @adrianlyjak in #9828
• Revert "Retain schema field ordering for google gemini and vertex" by @krrishdholakia in #10038
• Add aggregate team based usage logging by @krrishdholakia in #10039
• [UI Polish] UI fixes for cache control injection settings by @ishaan-jaff in #10031
• [UI] Bug Fix - Show created_at and updated_at for Users Page by @ishaan-jaff in #10033
• [Feat - Cost Tracking improvement] Track prompt caching metrics in DailyUserSpendTransactions by @ishaan-jaff in #10029
• Fix gcs pub sub logging with env var GCS_PROJECT_ID by @krrishdholakia in #10042
• Add property ordering for vertex ai schema (#9828) + Fix combining multiple tool calls by @krrishdholakia in #10040
• [Docs] Auto prompt caching by @ishaan-jaff in #10044
• Add litellm call id passing to Aim guardrails on pre and post-hooks calls by @hxmichael in #10021
• /utils/token_counter: get model_info from deployment directly by @chaofuyang in #10047
• [Bug Fix] Azure Blob Storage fixes by @ishaan-jaff in #10059
• build(deps): bump http-proxy-middleware from 2.0.7 to 2.0.9 in /docs/my-website by @dependabot in #10064
• fix(stream_chunk_builder_utils.py): don't set index on modelresponse by @krrishdholakia in #10063
• fix(llm_http_handler.py): fix fake streaming by @krrishdholakia in #10061
• Add aggregate spend by tag by @krrishdholakia in #10071
• Add OpenAI o3 & o4-mini by @PeterDaveHello in #10065
• Add new /tag/daily/activity endpoint + Add tag dashboard to UI by @krrishdholakia in #10073
• Add team based usage dashboard at 1m+ spend logs (+ new /team/daily/activity API) by @krrishdholakia in #10081
• [Feat SSO] Add LiteLLM SCIM Integration for Team and User management by @ishaan-jaff in #10072
• Virtual Keys: Filter by key alias (#10035) by @ishaan-jaff in #10085
• Add new /vertex_ai/discovery route - enables calling AgentBuilder API routes by @krrishdholakia in #10084
• fix(o_series_transformation.py): correctly map o4 to openai o_series … by @krrishdholakia in #10079
• [Feat] Unified Responses API - Add Azure Responses API support by @ishaan-jaff in #10116
• UI: Make columns resizable/hideable in Models table by @msabramo in #10119
• Remove unnecessary package*.json files by @msabramo in #10075
• Add Gemini Flash 2.5 Preview Model Price and Context Window by @drmingler in #10125
• test: update tests to new deployment model by @krrishdholakia in #10142
• [Feat] Support for all litellm providers on Responses API (works with Codex) - Anthropic, Bedrock API, VertexAI, Ollama by @ishaan-jaff in #10132
• fix(litellm-proxy-extras/utils.py): prisma migrate improvements: hand… by @krrishdholakia in #10138
• Litellm dev 04 18 2025 p2 by @krrishdholakia in #10157
• Gemini-2.5-flash - support reasoning cost calc + return reasoning content by @krrishdholakia in #10141
• Handle fireworks ai tool calling response by @krrishdholakia in #10130
• Support 'file' message type for VLLM video url's + Anthropic redacted message thinking support by @krrishdholakia in #10129
• fix(triton/completion/transformation.py): remove bad_words / stop wor… by @krrishdholakia in #10163
• Update model_prices_and_context_window_backup.json by @Classic298 in #10122
• to get API key from environment viarble of WATSONX_APIKEY by @ongkhaiwei in #10131
• test(utils.py): handle scenario where text tokens + reasoning tokens … by @krrishdholakia in #10165

New Contributors

• @Eoous made their first contribution in #9964
• @mrlorentx made their first contribution in #10006
• @hxmichael made their first contribution in #10021
• @chaofuyang made their first contribution in #10047
• @drmingler made their first contribution in #10125
• @Classic298 made their first contribution in #10122
• @ongkhaiwei made their first contribution in #10131

Docker Run LiteLLM Proxy

docker run \
-e STORE_MODEL_IN_DB=True \
-p 4000:4000 \
ghcr.io/berriai/litellm:main-v1.67.0-stable

Full Changelog: v1.66.0-stable...v1.67.0-stable