Comment by peder

Did you have any Anthropic vs OpenAI specification issues with Claude Code? I have been using mlx_vlm and vMLX and I get 400 Bad Request errors from Claude Code. Presumably you're not seeing those issues with llama-server ?