v1.85.3 - Observability, Budget & Rate-Limit Fixes
Deploy this version​
- Docker
- Pip
docker run \
-e STORE_MODEL_IN_DB=True \
-p 4000:4000 \
docker.litellm.ai/berriai/litellm:1.85.3
pip install litellm==1.85.3
v1.85.3 is a patch release on top of v1.85.2. It cherry-picks fixes for duplicate Claude Code traces, Bearer-prefix hashing, budget-reset writes, and two flag-leak corrections in the rate limiter and the provider request body.
What's Changed​
- fix(logging): stop duplicate Claude Code traces (internal copy of #29089) - PR #29311
- fix(proxy): normalize the Bearer prefix in the safe-hash helper - PR #29343
- fix(budget): reset_budget writes only
{spend, budget_reset_at}and no longer pre-zeroes the counter - PR #29358 - fix(rate-limit): stop the v3 limiter from leaking internal stash to the provider body - PR #27913
- fix(proxy): stop the
use_chat_completions_apiflag from leaking into the provider request body - PR #29447
Full Changelog​
https://github.com/BerriAI/litellm/compare/v1.85.2...v1.85.3