ollama - 💡(How to fix) Fix Loading gemma4:26b-nvfp4 raises "Error: 500: mlx runner failed: Error: unsupported architecture: Gemma4ForConditionalGeneration" [3 comments, 4 participants]

Error: 500 Internal Server Error: mlx runner failed: Error: unsupported architecture: Gemma4ForConditionalGeneration (exit: exit status 1)

TL;DR

The issue is likely due to the Gemma4ForConditionalGeneration model not being compatible with the Apple M3 architecture, and a workaround may involve using a different model or version.

Guidance

• The error message "unsupported architecture" suggests that the Gemma4ForConditionalGeneration model is not compatible with the Apple M3 chip, which is used in the MacBook.
• Try running the model on a different device or environment that supports the required architecture.
• Compare the compatibility requirements of the Gemma4ForConditionalGeneration model with the qwen3.5 model, which is reported to work without issues.
• Check the Ollama version 0.20.7 documentation to see if there are any known issues or workarounds for running models on Apple M3 devices.

Notes

The issue may be specific to the combination of the Gemma4ForConditionalGeneration model, Ollama version 0.20.7, and the Apple M3 architecture, and more information may be needed to determine a definitive solution.

Recommendation

Apply workaround: Try using a different model or version that is compatible with the Apple M3 architecture, as the error message suggests that the current model is not supported.