Question 1

Can NVIDIA V100 32GB run Mistral Small 3.2 24B?

Accepted Answer

Yes, NVIDIA V100 32GB can run Mistral Small 3.2 24B with a C grade (Runs well). Expected decode speed: 44.3 tok/s.

Question 2

How much VRAM does Mistral Small 3.2 24B need?

Accepted Answer

Mistral Small 3.2 24B (24B parameters) requires approximately 22.8 GB of memory with Q4_K_M quantization.

Question 3

What is the best quantization for Mistral Small 3.2 24B?

Accepted Answer

The recommended quantization for Mistral Small 3.2 24B is Q4_K_M, which balances quality and memory efficiency.

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	C	Tight fit	41.2 tok/s	6837 ms	39K
Chat	C	Runs well	41.2 tok/s	2564 ms	12K
Coding	C	Runs well	44.3 tok/s	4374 ms	22K
RAG	C	Tight fit	44.3 tok/s	7953 ms	39K
Reasoning	C	Runs well	41.2 tok/s	5555 ms	22K

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	9.4 GB	Low	D35
Q3_K_S	3	11.8 GB	Low	D37
NVFP4	4

Can NVIDIA V100 32GB run Mistral Small 3.2 24B?

Memory breakdown