Atlas

Models

← All models

Devstral Small 2

Mistral AI · 2025-12-09 · 24B parameters

Devstral Small 2 is Mistral AI's December 2025, 24B-class open-weight multimodal agentic-coding model, released alongside the 123B Devstral 2. It accepts text and image inputs and is designed for tool-driven codebase exploration, multi-file editing, and software-engineering agents. The checkpoint supports a 262,144-token context, is published in FP8 form, and carries the Apache-2.0 weight license.

Benchmark scores

BenchmarkScore
AA-Briefcase Elo496
AA-Briefcase v1.1 Elo496
AA-LCR24.0
AA-LCR v1.128.0
AA-Omniscience Index-57.7
AIME 2025 (Artificial Analysis avg@10)34.3
Artificial Analysis Agentic Index10.5
Artificial Analysis Coding Index29.3
Artificial Analysis Intelligence Index17.4
Artificial Analysis Intelligence Index v4.38.4
Artificial Analysis Intelligence Index v4.3.27.5
Artificial Analysis Omniscience Accuracy15.8
Artificial Analysis Omniscience Hallucination Rate87.3
Artificial Analysis Openness Index27.8
Artificial Analysis Progress Index v1185
AutomationBench-AA 1.0.63.1
Capability130.9
CritPt (avg@5)0.0
GDP.pdf (Artificial Analysis, All-pass)1.0
GDP.pdf (Artificial Analysis, Mean Criteria)38.8
GDPval-AA v2680
GDPval-AA v2 (normalized)11.7
GDPval-AA v2.1536
Global MMLU Lite69.2
GPQA Diamond (Artificial Analysis avg@5)53.2
Humanity's Last Exam (Text-Only)3.5
IFBench (Artificial Analysis avg@5)31.2
IOI2.5
IOI - IOI 20243.7
IOI - IOI 20251.3
Loading Atlas data…