ollama - 💡(How to fix) Fix MLX runner failed with qwen3.6:35b-a3b-coding-bf16 format=json [2 comments, 2 participants]

{"error":"mlx runner failed: time=2026-04-26T21:12:59.383+09:00 level=INFO source=pipeline.go:129 msg="Prompt processing progress" processed=10 total=11"}

TL;DR

The issue with the curl request to the API endpoint may be related to the model or input data, and checking the model compatibility or input format could be the first step towards a solution.

Guidance

• Verify that the model "qwen3.6:35b-a3b-coding-bf16" is compatible with the current version of Ollama (0.21.2) and the Apple GPU.
• Check the input data format, specifically the "messages" field, to ensure it matches the expected format for the API endpoint.
• Investigate the "mlx runner failed" error message in the log output, which may indicate a problem with the pipeline or model execution.
• Consider checking the Ollama documentation for any specific requirements or restrictions on model usage or input data.

Notes

The provided log output does not contain a clear error message, and the issue may require further investigation or additional information to determine the root cause.

Recommendation

Apply workaround: Check model compatibility and input data format, as these are the most likely causes of the issue, and adjusting these may resolve the problem without requiring an upgrade or significant changes.