Mistral Small 3.1

Efficient open-weight Mistral-family model often considered for practical local chat and app workloads.

Mistral AI · Mistral

Editorial review

Reviewed byOpenSourcesAI EditorialLast updatedAugust 2026SourcesMistral model card

Model checkpoints, context windows, provider support, local runtime compatibility, and license terms can change quickly. Verify the exact model card before production or commercial use.

Best for

Builders who want a smaller capable model with permissive licensing and good local runtime support.

Who should use it

Builders who want a smaller capable model with permissive licensing and good local runtime support.
Builders who want local or self-hosted testing options.

Common workflows

Efficient chat, local apps, multilingual work
chat workflows
efficient workflows
local workflows

Deployment and hardware notes

24B-class models are realistic on higher-end local systems with quantization.

License and usage notes

Apache 2.0. Open weights. Verify the exact model card and license terms for the checkpoint or hosted provider you use.

Strengths

Open weights model option for Mistral workflows.
Builders who want a smaller capable model with permissive licensing and good local runtime support.
A practical local candidate when quantized; still test memory use and context length on your hardware.
Tracked as Practical local in the OpenSourcesAI model directory.

Limitations

May not match larger MoE models on hard reasoning; compare on your own prompts before production.
24B-class models are realistic on higher-end local systems with quantization.
Context window and limits: 131,072 tokens.
Verify the exact model card, provider docs, license, and serving support before production use.

Local workflow notes

A practical local candidate when quantized; still test memory use and context length on your hardware.

Local runtimes: Ollama, LM Studio, llama.cpp, vLLM

Platforms: Windows, macOS, Linux

Will Mistral Small 3.1 run on your machine?

Mistral Small 3.1 is 24B parameters and needs 15.9 GB of VRAM at Q4_K_M — 14.4 GB of weights plus 1.5 GB of runtime overhead for the inference server itself.

VRAM by quantization

Quantization	Weights	Needs (with overhead)	Quality
Q4_K_M	14.4 GB	15.9 GB	good
Q8_0	25.5 GB	27 GB	high
FP16	48 GB	49.5 GB	reference

Fit on common hardware at Q4_K_M

Hardware	Memory the model can use	System RAM	Verdict
CPU Only	None (CPU only)	16 GB	Too large
RTX 4060 Laptop	8 GB	16 GB	Too large
RTX 3060 (12GB)	12 GB	32 GB	CPU offload
RTX 4060 Ti (16GB)	16 GB	32 GB	Tight
RTX 3090	24 GB	64 GB	Comfortable
Apple Silicon (Unified Memory) 36 GB	27 GB of 36 GB	36 GB	Comfortable
RTX 5090	32 GB	64 GB	Comfortable

Comfortable means VRAM clears the requirement by 2 GB or more. Tight means it covers the requirement with no margin. CPU offload means the model does not fit in VRAM but system RAM is at least 1.6× the weights, so it will run at reduced speed — expect roughly 1–5 tokens per second. Figures are weights plus a fixed runtime overhead and exclude KV-cache growth, which scales with context length.

Apple Silicon shares one pool of memory between the system and the GPU, so a model cannot use all of it. These rows apply the same 75% usable fraction the Compatibility Checker uses, which is why a 36 GB Mac is graded on less than 36 GB.

Check Mistral Small 3.1 against your exact hardware How these verdicts are calculated