BerriAI/litellm v1.108.9 on GitHub

Verify Docker Image Signature

All LiteLLM Docker images are signed with cosign. This release is signed with the key introduced in commit e82a3c6.

Verify using the pinned commit hash (recommended):

A commit hash is cryptographically immutable, so this is the strongest way to ensure you are using the published signing key:

cosign verify \
  --key https://raw.githubusercontent.com/BerriAI/litellm/e82a3c646908ba0a6ee0beddc4310d7a6cac4065/cosign.pub \
  ghcr.io/berriai/litellm:v1.108.9

Verify using the release tag (convenience):

Tags are protected in this repository and resolve to the same key. This option is easier to read but relies on tag protection rules:

cosign verify \
  --key https://raw.githubusercontent.com/BerriAI/litellm/v1.108.9/cosign.pub \
  ghcr.io/berriai/litellm:v1.108.9

Expected output:

The following checks were performed on each of these signatures:
  - The cosign claims were validated
  - The signatures were verified against the specified public key

What's Changed

fix(responses): backport mixed reasoning and answer preservation to lts/1.108.x by @mateo-berri in #51356
fix(bedrock): backport Converse s3:// media sources to lts/1.108.x by @mateo-berri in #50401
fix(azure_ai): send reasoning_effort to Foundry Kimi K3 and keep its reasoning_content on lts/1.108.x by @tavren27 in #51748
feat(vertex_ai): backport caller identity labels on Vertex requests to lts/1.108.x by @mika-lt in #51786
fix(responses): backport cache_control breakpoints on the chat-to-Responses bridge to lts/1.108.x by @norkelbyte in #51796
fix(azure): backport Foundry project and resource /openai/v1 Responses bases to lts/1.108.x by @mika-lt in #51769
feat(decisions): backport Vercel AI Gateway decision models to lts/1.108.x (#50642) by @quietjules in #51250
fix(bedrock): backport OpenAI vector store search filter mapping to lts/1.108.x by @mika-lt in #51292
chore: bump litellm to 1.108.9 on lts/1.108.x by @yuneng-berri in #51853

Full Changelog: v1.108.8...v1.108.9