●Change8

v1.77.7.dev-gemma

📦 litellmView on GitHub →
✨ 2 features🐛 9 fixes

Summary

This release focuses on stability and feature enhancements, including fixes for costing, rate limiting, and Anthropic tool calls, alongside the addition of EnkryptAI Guardrails.

✨ New Features

  • Added EnkryptAI Guardrails integration to LiteLLM.
  • Added pricing details for new Azure AI models.

🐛 Bug Fixes

  • Fixed litellm_param based costing.
  • Fixed parallel tool calls in the Anthropic passthrough adapter.
  • Redacted AWS credentials when redact_user_api_key_info is enabled.
  • Fixed dynamic Rate limiter v3 by inserting litellm_model_saturation.
  • Fixed sessions not being shared.
  • Prevented DB from accidentally overriding config file values if they are empty in DB.
  • Temporarily relaxed ResponsesAPIResponse parsing to support custom backends (e.g., vLLM).
  • Fixed OpenRouter cache_control to only apply to the last content block.
  • Removed panic from a hot path.