Hacker News new | ask | show | jobs
by anuarsh 264 days ago
I haven't tested on Apple machines yet, but gpt-oss and qwen3-next should work I assume. Llama3 versions use cuda specific loading logic for speed boost, so it won't work for sure