Earlier, I wrote about Qwen 3.8 being good but slow. Since then, I have been trying to figure out how to boost the speed of running Qwen 3.8 on my 64GB M2 Ultra Mac Studio.

What is working for me is mlx-serve. This is kinda Ollama-ish, in that it has an inference server with an associated GUI. Overall, it does quite a bit more than Ollama does. But the biggest thing is that it runs Qwen 3.8 much better than Ollama does. My estimate is that at agentic coding, Qwen 3.8 via OpenCode is somewhere around 25% of the speed of Claude Sonnet via Claude Code, with a bit lower quality.

On the one hand, that means that Sonnet is much faster and gives higher-quality results. On the other hand, Qwen 3.8 is a local model:

• No per-token fees (directly or in the form of consuming subscription quotas)
• No contributing to data center loads for inference work

Right now, I am having Sonnet review Qwen's work, but so far, Qwen has reliably delivered what it was supposed to do. And this is about a 4x speed increase from what I had with my Ollama-based Qwen 3.8 setup.

Part of that speed increase may be from better MLX support in mlx-serve than in Ollama. The bigger thing might be the support that mlx-serve has for MTP (Multi-Token Prediction). MTP is a form of "speculative decoding", where lesser models do the heavy lifting and the top-tier model chooses among their outputs. Qwen 3.6 and 3.8 offer MTP, and mlx-serve lets me toggle that on.

There are clear limits. Having Qwen 3.8 do agentic coding consumes 99% of the GPU of my Mac Studio, and that is with just a single conversation. Agent swarms require a fleet of hardware, not just a single Mac.

Still, this is more than good enough for lightweight tasks and lets me offload some of my development from frontier models and their data centers.