How much VRAM does Mistral Small 3.2 24B need?

At Q4_K_M with 8K context: 14.7GB weights + 1.3GB KV cache + 1.5GB runtime overhead = 17.5GB total. Mac Studio M3 Ultra (96GB) offers 72.0GB usable VRAM, so the verdict is: Comfortable.

What is the best quantization for Mistral Small 3.2 24B on Mac Studio M3 Ultra (96GB)?

FP16 — the highest tier that still fits within 72.0GB usable VRAM at 8K context. Lower tiers (Q3/Q2) fit too but cost noticeable quality.

How fast does Mistral Small 3.2 24B run on Mac Studio M3 Ultra (96GB)?

About 13 tokens/s at FP16 (theoretical estimate, ±30% in real-world use).

Can Mac Studio M3 Ultra (96GB) run Mistral Small 3.2 24B?

✅ Comfortable

Mistral Small 3.2 24B @ Q4_K_M · 8K context

Needed

17.5 GB

Usable

72.0 GB

VRAM breakdown

Weights (Q4_K_M)	14.7 GB
KV cache (8K context)	1.3 GB
Runtime overhead	1.5 GB
Total	17.5 GB

Computed at 8K context with fp16 KV cache. Longer contexts need more VRAM — use the VRAM calculator for other settings.

Recommended quantization

FP16

Estimated generation speed

~13 tok/s @ FP16

Theoretical estimates based on published architecture data and measured GGUF sizes; real-world speed varies ±30%.

Cheaper GPUs that run it

Smaller models this GPU runs well

Other GPUs that run Mistral Small 3.2 24B

FAQ

How much VRAM does Mistral Small 3.2 24B need?: At Q4_K_M with 8K context: 14.7GB weights + 1.3GB KV cache + 1.5GB runtime overhead = 17.5GB total. Mac Studio M3 Ultra (96GB) offers 72.0GB usable VRAM, so the verdict is: Comfortable.
What is the best quantization for Mistral Small 3.2 24B on Mac Studio M3 Ultra (96GB)?: FP16 — the highest tier that still fits within 72.0GB usable VRAM at 8K context. Lower tiers (Q3/Q2) fit too but cost noticeable quality.
How fast does Mistral Small 3.2 24B run on Mac Studio M3 Ultra (96GB)?: About 13 tokens/s at FP16 (theoretical estimate, ±30% in real-world use).

Check another combination in the GPU Checker →

Data verified 2026-09-01