Help me choose
Three questions. Answers are filtered to what actually fits your M4 Pro with 48 GB.
Start here
Mixtral 8x7B Instruct v0.1FITS · 11.1 GB FREE
One of the earliest MoE models to run well on Apple Silicon, needing a 32GB Mac for 4-bit despite having only 12.9B active parameters per token. It still runs faster than a dense 46B model would, but newer MoE designs like Qwen3-30B-A3B have mostly superseded it on quality.
46.7B · q4 · 29.3 tok/s est.
mlx_lm.chat --model mlx-community/Mixtral-8x7B-Instruct-v0.1-4bitRunners-up
- Qwen2.5-VL 32B Instructq4 · 11.5 tok/s
- Qwen3 32Bq6 · 8.0 tok/s
- Qwen2.5 32B Instructq4 · 11.6 tok/s
- Qwen2.5 Coder 32B Instructq4 · 11.6 tok/s
Buying rather than choosing? 20 chips are compared on the hardware page.