Compatibility check
Can I run Canary Qwen 2.5b on Apple M3 Ultra (80-core GPU)?
Yes — it runs.best quant:
Decode speed
364 tok/sest
Usable context
262K
Memory
512 GB
Fits in memory364 tok/sest · 262K ctx
Fits in memory310 tok/sest · 262K ctx
Fits in memory208 tok/sest · 262K ctx