v1.81.13-nightly
📦 litellmView on GitHub →
✨ 3 features🐛 13 fixes🔧 6 symbols
Summary
This release focuses heavily on infrastructure improvements, particularly stabilizing tests across multiple providers and fixing various bugs related to API parameters, SSO, and managed batches. New UI features include Access Group selection and Content Filter management.
Migration Steps
- If you were relying on the specific value of the reasoning parameter, update your code to use None instead of the previous Reasoning() object.
- If using Vertex AI rerank, check tests related to reasoning_effort as the expected format might have changed to dict.
✨ New Features
- UI: Added Access Group Selector to Create and Edit Flow.
- Added routing logic based on whether reasoning is supported.
- UI: Added Content Filters management for editing/viewing categories and 1-click add categories.
🐛 Bug Fixes
- Preserved key_alias and team_id metadata in /user/daily/activity/aggregated after key deletion or regeneration.
- Fixed reasoning parameter usage by using None instead of Reasoning().
- Removed unused Reasoning import from transformation.py.
- Fixed ChatGPT Codex unsupported parameters issue.
- Fixed Converse Anthropic usage object to align with v1/messages specs.
- Applied miscellaneous bug fixes for Managed batches.
- Clamped thinking.budget_tokens to a minimum of 1024 for Bedrock.
- Fixed virtual key grace period retrieval from environment variables/UI.
- Fixed SSO PKCE support failure in multi-pod Kubernetes deployments.
- Excluded tool parameters for models that do not support function calling.
- Fixed handling of missing DATABASE_URL in append_query_params for proxy.
- Reverted StreamableHTTPSessionManager to stateless mode for MCP.
- Fixed OCI Grok output pricing.