Devstral Small 2
Mistral · 1 configuration
Intelligence7.5
Input / 1M$0.10
Output / 1M$0.30
Context256K
Tariff: Pioneer · Source: models-dev · USD per million tokens
Thinking configuration
What does more thinking buy you?
Compare this model’s measured thinking configurations.
Scroll the chart ↔ · tap a point for details
Efficient frontier: best measured score at each cost1 measured configurations · 0 without both values
Selected: thinking default · Click a point to switch configuration
Compact multimodal coding model for repository exploration, file editing, and software agents
Intelligence IndexArtificial Analysis composite index, not a percentage. Different versions of the index are not directly comparable.
7.5 pointsGPQA DiamondGraduate-level scientific questions. Measures specialist reasoning rather than web-search quality.
53.2 %Humanity’s Last ExamDifficult questions across many disciplines, compared under the same evaluation setup.
3.5 %SciCodeCode generation for scientific research problems.
32.4 %Terminal-Bench 2.1Agent tasks in a terminal environment. Results also depend on the evaluation harness.
29.6 %τ²-BenchTool use and interaction in assistance scenarios. Does not replace an evaluation of your integration.
23.4 %IFBenchVerifiable instruction following. A proxy for precision rather than a benchmark of creative style.
31.2 %MMMU-ProMultimodal understanding of academic problems that include images.
44.6 %Omniscience accuracyFactual accuracy in Artificial Analysis's Omniscience evaluation.
15.8 %Specifications from the catalog
Inputtext, image
Outputtext
Tool callingYes
Structured outputNot available
SpeedNot available
First tokenNot available
Maximum output131.1K
Catalog listing date2025-12-09