v1.85.7 - Stream Cost Recovery & Cache-Control Cap
Deploy this version​
- Docker
- Pip
docker run \
-e STORE_MODEL_IN_DB=True \
-p 4000:4000 \
docker.litellm.ai/berriai/litellm:1.85.7
pip install litellm==1.85.7
v1.85.7 is a patch release on top of v1.85.6. It backports cost-tracking recovery for interrupted Anthropic streams, caps Anthropic cache_control injection at the 4-block API limit, and pulls in OpenSSL and OSV-flagged dependency bumps for CVE coverage. It also hardens usage parsing by coercing a dict server_tool_use into a typed ServerToolUse object.
What's Changed​
- fix(integrations): cap Anthropic cache_control injection at 4 blocks - PR #30480
- fix(passthrough): recover output tokens for interrupted anthropic streams - PR #30787
- fix(proxy): record partial spend on the failure row for interrupted streams - PR #30788
- fix(passthrough,streaming): recover cost on interrupted and agentic Anthropic streams - PR #31035
- fix(deps): bump osv-flagged dependencies to clear known CVEs - PR #31122
- fix(docker): bump wolfi-base digest to patch openssl CVE-2026-34182 - PR #31133
Full Changelog​
https://github.com/BerriAI/litellm/compare/v1.85.6...v1.85.7