Question 1

Can I run Mistral Small 24B on a Apple M1?

Accepted Answer

Mostly — Apple M1 runs Mistral Small 24B (Q4_K_M) with partial GPU offload. Expect slower speeds than a fully fitting card. Partial GPU offload — 16 GB VRAM is above the 16 GB minimum but below the 20 GB recommendation. Some layers will spill to RAM.

Question 2

What quantization of Mistral Small 24B should I use on a Apple M1?

Accepted Answer

For 16 GB VRAM on the Apple M1, the Q4_K_M variant is the best fit. Estimated ~2.7 tokens/sec on the Q4_K_M quantization.

Question 3

How fast does Mistral Small 24B run on Apple M1?

Accepted Answer

Roughly 2.7 tokens/sec for Q4_K_M. Real speed depends on context length, backend (Ollama, llama.cpp, LM Studio), and KV cache size.

Question 4

What if Apple M1 is not enough for Mistral Small 24B?

Accepted Answer

Try a smaller quantization variant or run on CPU + RAM with the trade-off of slower generation.

Quantization	File Size	Min VRAM	Rec VRAM	Context	Verdict	Estimated tok/s
Q4_K_MBest fit	14 GB	16 GB	20 GB	8K / 32K	Partial GPU	~2.7
Q8_0	25 GB	27 GB	32 GB	8K / 32K	Hybrid CPU+GPU	~1

Can I Run Mistral Small 24B on Apple M1?

Share this matchup

Every Mistral Small 24B quantization on Apple M1