litellm - 💡(How to fix) Fix [Feature]: Official support for Gemma-4 with tool calling via Claude Code [1 participants]

TL;DR

Add native support for the Gemma-4 model family in the proxy feature of LiteLLM to enable compatibility with Claude Code on Google AI Studio's free tier.

Guidance

• Investigate the current implementation of the proxy feature in LiteLLM to understand how it interacts with different models.
• Research the specific requirements of the Gemma-4 model family (gemma-4-31b-it and gemma-4-26b-a4b-it) to determine what changes are needed for native support.
• Consider the limitations and capabilities of Google AI Studio's free tier to ensure that the implementation is optimized for this environment.
• Evaluate the compatibility of Claude Code with the proposed changes to ensure a seamless user experience.

Notes

The implementation details of the Gemma-4 model family and the proxy feature in LiteLLM are not provided, so a more specific solution cannot be offered.

Recommendation

Apply workaround: Implement a custom proxy solution that supports the Gemma-4 model family, allowing users to utilize Claude Code on Google AI Studio's free tier until native support is added.