Mistral Small

Cost-efficient model for focused tasks

Mistral Small 3 offers a strong capability-to-cost ratio, designed for tasks requiring focused intelligence: Q&A, summarisation, classification, and lightweight agentic workflows.

Try Mistral Small

Specifications

Context window32k tokens

Input price$0.100 / 1M tokens

Output price$0.300 / 1M tokens

Credits per query2 cr

Released2025-03

LicenseApache 2.0

Features

Tool useJSON modeFunction calling

Benchmark Scores

What do these mean?

Sourced from official model cards and academic papers. Higher is better.

GPQA Diamond

PhD-level science & reasoning

—

AIME 2024

Mathematical olympiad problems

—

Arena Elo

Chatbot Arena human preference ranking

—

HumanEval

Python coding — pass@1 rate

—

MATH

Competition-level mathematics

—

MMLU

General knowledge across 57 academic domains

77%

MT-Bench

Multi-turn instruction following (out of 10)

—

SWE-Bench

Real-world GitHub issue resolution rate

—

Run locally

Download and run this model on your own hardware — no API key or internet required after setup.

Min VRAM13 GB

Rec. VRAM16 GB

Best VRAM26 GB

Parameters24B

Speed~13 tok/s

Context32K tokens

Ollama

Terminal

Run this command in your terminal. Ollama downloads and manages the model locally.

ollama pull mistral-small

LM Studio

Desktop app · no terminal needed

Hugging Face

Raw weights · any runtime

More from Mistral

Mistral Large3 cr

Mistral's flagship reasoning and multilingual model

Magistral Medium1 cr

Mistral's frontier reasoning model

Pixtral Large1 cr

Mistral's frontier vision and language model