Devstral Small 2 24B

Overview

Mistral AI's 24B coding specialist co-developed with All Hands AI, scoring 72.2% on SWE-Bench under Apache 2.0. Fits on a single RTX 4090.

When to pick this model

• Single-GPU coding agents on a 4090
• Repository-scale refactoring up to 256K tokens
• SWE-Bench-style autonomous coding tasks
• Apache-licensed commercial code tools
• European-lab-sourced coding infrastructure

VRAM requirements by quantization

Quantization	VRAM required
Q4_K_M (recommended)	14 GB
Q5_K_M	17 GB
Q8_0	26 GB
FP16 (no quantization)	48 GB

VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.

Published benchmark scores

Benchmark	Score
SWE-Bench	72.2

Scores published by the model author or aggregated from public leaderboards. Re-measured monthly by our editorial team.

Strengths

• 72.2% SWE-Bench in a 24B dense model
• Runs comfortably on a single RTX 4090
• 256K context for whole-repo work
• Apache 2.0 license
• Co-developed with All Hands AI for agent workloads

Limitations

• No vision capability
• Specialized for code, weaker as a general assistant

Architecture & training

Architecture: Dense 24B · Mistral base · 256k ctx · code post-trained

Training: Co-developed with All Hands AI.

Quick start

ollama run devstral-small2:24b

Or use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.