Mistral-Small-Instruct-2409 — benchmark results

Mistral's 22B dense instruct model with function calling and a 32K context, under the Mistral Research License (September 2024). Provider: Mistral. Released 2024-09-17. Access: Open.

Unified ELO 1539 ± 16, rank #650 of 1776 rated models, from 38 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
MT-Bench PL - Humanities10Judge Score (0-10)94.9
Polish EQ-Bench72.85EQ-Bench Score91.1
HREF42.87Average HREF Score (%)87.9
MT-Bench PL - Reasoning7.9Judge Score (0-10)82.7
MT-Bench PL - STEM9.65Judge Score (0-10)82.7
Open LLM Leaderboard - GPQA11.07Score81.7
MT-Bench PL - Coding7.1Judge Score (0-10)81.6
Open CoT - LogiQA 212.66CoT Gain (%)81.3
Open CoT - LSAT Logical Reasoning17.06CoT Gain (%)80.2
Open LLM Leaderboard - BBH40.56Score79.5
Open LLM Leaderboard - IFEval66.7Score78.7
Open PL LLM - Multiple Choice57.99Average Multiple-Choice Score (%)78.7

Interactive version: theaggregate.ai/model?slug=mistral-small-instruct-2409 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.