What Happened
LiteLLM published updates on three release lines. The 1.110.0-rc.3 candidate backports router request-field validation, fixes, and dependency updates from rc.2. The 1.107.0-rc.4 candidate adds lite credentials connect for linking per-user GitHub Copilot credentials without the Admin UI and updates locked Docker dependencies, including aiohttp, ddtrace, and pypdf. Its release notes identify 1.109.1 as the current stable release at that time. [1] [2]
The 1.108.9 long-term-support update backports fixes for Responses reasoning, cache-control breakpoints, Bedrock media and search filters, and Azure AI/Foundry routing and reasoning. It also adds Vertex AI caller-identity labels and Vercel AI Gateway decision models. [3]
A Streamlit nightly build, 1.67.1.dev20261113, was listed without release notes. Its changes and compatibility impact cannot be determined from the available information. [4] None of these notes identifies a confirmed breaking change; router request-field validation in 1.110.0-rc.3 nevertheless warrants compatibility testing. [1]
Why It Matters to Businesses
A LiteLLM upgrade can change behavior at the shared gateway used by multiple applications. The 1.108.9 fixes are relevant to teams using the affected providers and features, while the release candidates should be evaluated against a specific need rather than treated as routine production upgrades. [1] [2] [3]
The LiteLLM Docker releases state that their images are signed with a replacement cosign key. Signature verification offers a way to check the published image’s signing claims, but does not establish that the code is defect-free. The artifact and signature claims in the release notes have not been independently verified here. [1] [2] [3]
Kimbodo Engineering Perspective
Choose the release line before choosing the version. If an application needs a fix backported to 1.108.9, test that long-term-support update against its existing gateway configuration. Use a release candidate only when its particular feature or fix justifies the additional validation. Do not infer Streamlit upgrade safety from a nightly version number alone. [1] [2] [3] [4]
How We Would Implement It
- Inventory deployed LiteLLM versions, provider integrations, request fields, caching behavior, and credential workflows; map each dependency to the relevant change. [1] [2] [3]
- Fetch the intended Docker image, verify its cosign signature with the public key pinned to commit e82a3c646908ba0a6ee0beddc4310d7a6cac4065, and deploy by image digest rather than a mutable tag. [1] [3]
- Run staging regression tests for routing and request validation, Responses reasoning, cache controls, and each affected provider path in use. Test lite credentials connect only if adopting the 1.107 release-candidate line. [1] [2] [3]
- Canary the upgrade with gateway error rates, latency, and provider-specific failures visible; retain the previous image digest for rollback.
Risks, Costs and Security
Validation costs rise when organizations maintain several LiteLLM release lines: each needs its own artifact, dependency, and integration tests. The dependency updates in 1.107.0-rc.4 and request validation in 1.110.0-rc.3 are reasons to test actual workloads, not evidence of a documented breaking change. [1] [2]
Use the pinned public key for signature verification rather than relying solely on a release tag. Protect credential-linking workflows with the same access controls and audit practices as other user credentials. Defer the Streamlit nightly in production until its changes can be reviewed and tested. [1] [2] [3] [4]
Where Kimbodo Comes In
Kimbodo builds and operates this in production for businesses — see our AI Application Development practice, or Estimate My AI Application.
Sources
- [1] v1.110.0-rc.3
- [2] v1.107.0-rc.4
- [3] v1.108.9
- [4] 1.67.1.dev20261113