What's Changed
- fix(proxy): recover from prisma-query-engine zombie process by @hcavarsan in #21899
- fix: apply custom video pricing from deployment model_info by @AtharvaJaiswal005 in #21923
- fix: team assignment fails for keys with special model names by @AtharvaJaiswal005 in #21919
- Revert "fix: team assignment fails for keys with special model names" by @krrishdholakia in #21977
- fix(ollama): thread api_base to get_model_info + graceful fallback by @Chesars in #21970
- fix(anthropic): sanitize tool_use IDs in convert_to_anthropic_tool_invoke by @Canvinus in #21964
- fix(aiohttp): only set enable_cleanup_closed when required by @SolitudePy in #21897
- [Doc] replaced azure openai key with mock key by @shivamrawat1 in #21997
- fix: atomic RPM rate limiting in model rate limit check by @jquinter in #21957
- Revert PR #21957: atomic RPM rate limiting by @jquinter in #22001
- litellm oss staging 02 23 2026 by @krrishdholakia in #21940
- Fix: skip health check for MCP integration with passthrough token auth by @Sameerlite in #21982
- fix: Missing OAuth session state by @Sameerlite in #21992
- Fix: Transport Type for OpenAPI Spec on UI by @Sameerlite in #22005
New Contributors
Full Changelog: litellm-CVE-v1.81.15...CVEv1.81.5