---
title: "Devstral Medium Benchmarks - RankedAGI"
description: "Detailed benchmark and metadata record for Devstral Medium."
source: "https://rankedagi.com/models/devstral-medium"
---

# Devstral Medium

| Field | Value |
| --- | --- |
| Organization | Mistral |
| License | Open Source |
| Version | Latest |
| Released | 2025-07-10 |
| Input cost per million tokens | $0.4 |
| Output cost per million tokens | $2 |
| Last updated | 2026-04-27T17:01:29.321Z |

## Benchmarks

| Benchmark | Category | Value | Description | Source |
| --- | --- | --- | --- | --- |
| RankedAGI Coding | coding | 41.4% | RankedAGI Coding Score |  |
| SWEBench Verified | coding | 61.6% | Agentic Coding | https://www.swebench.com/index.html |
| RankedAGI Agentic | Agents | 31.0% | RankedAGI Agentic Score |  |
| Code Arena | coding | 1091 | Code Arena Elo Score | https://arena.ai/leaderboard/code |
| LiveCodeBench v6 | coding | 33.7% | LiveCodeBench v6 - Evaluation of Large Language Models for Code |  |
| RankedAGI Reasoning | reasoning | 31.7% | RankedAGI Reasoning Score |  |
| Humanity's Last Exam | reasoning | 3.8% | Multidisciplinary Reasoning (no tools) |  |
| GPQA Diamond | reasoning | 49.2% | Generalized Prefix Question Answering Score (Reasoning) PhD Level Reasoning |  |
| AIME 2025 | math | 4.7% | AIME 2025 Competition Math |  |
| AIME 2024 | math | 6.7% | AIME 2024 Competition Math |  |
| MMLU Pro | knowledge | 70.8% | Graduate-level visual problem-solving |  |
| Instruction Following Evaluation | instruction-following | 29.9% | Instruction Following Evaluation |  |
| RankedAGI Math | math | 36.7% | RankedAGI Math Score |  |
| RankedAGI Overall | general | 39.3% | Overall RankedAGI score |  |
| GDPval-AA Elo | Agents | 684 | Office Tasks (Artificial Analysis) |  |
| Code DesignArena | Design | 1077 |  | https://www.designarena.ai/leaderboard/code |