---
title: "Mistral Small 3 24B Benchmarks - RankedAGI"
description: "Detailed benchmark and metadata record for Mistral Small 3."
source: "https://rankedagi.com/models/mistral-small-3"
---

# Mistral Small 3

| Field | Value |
| --- | --- |
| Organization | Mistral |
| License | Proprietary |
| Size | 24B |
| Version | Latest |
| Released | 2025-01-30 |
| Context window | 32K |
| Last updated | 2025-04-16 22:14:46.146909+00 |

## Benchmarks

| Benchmark | Category | Value | Description | Source |
| --- | --- | --- | --- | --- |
| RankedAGI Coding | coding | 30.8% | RankedAGI Coding Score |  |
| RankedAGI Agentic | Agents | 30.3% | RankedAGI Agentic Score |  |
| RankedAGI Reasoning | reasoning | 36.6% | RankedAGI Reasoning Score |  |
| GPQA Diamond | reasoning | 45.3% | Generalized Prefix Question Answering Score (Reasoning) PhD Level Reasoning |  |
| HumanEval | coding | 84.8% | HumanEval coding benchmark |  |
| NYT Connections Extended | reasoning | 8.9% | NYT Connections Extended Version | https://github.com/lechmazur/nyt-connections/ |
| MMLU Pro | knowledge | 66.3% | Graduate-level visual problem-solving |  |
| MMLU | knowledge | 81.0% | Image Reasoning |  |
| Hughes Hallucination Evaluation | safety | 3.1% | Computed using Vectara's Hallucination Evaluation Model, also known as HHEM. This evaluates how often an LLM introduces hallucinations when summarizing a document. | https://github.com/vectara/hallucination-leaderboard |
| Instruction Following Evaluation | instruction-following | 82.9% | Instruction Following Evaluation |  |
| RankedAGI Math | math | 46.5% | RankedAGI Math Score |  |
| RankedAGI Overall | general | 38.4% | Overall RankedAGI score |  |