Devstral Medium
Provider: Mistral • ID: devstral-medium
Detailed evaluation metrics and judge breakdown for Devstral Medium across all 4 launch categories.
Compare in Blind Arena
Pricing: In — • Out —
Composite Verdict Score/100
12.4
0
.0
Category Score Breakdown
Frontend UI12.7
Interactive dashboards and landing pages
Game Dev12.2
Canvas 2D physics and browser games
SVG Art12.5
Vector graphics and generative math art
Agentic Tasks12.0
Multi-step execution planning
Auditable Multi-Judge Dimension Scoring
Grades evaluated across 3 independent judge models using published rubrics.
Functionality— Executes without runtime JS exceptions
12.8
Craft— Clean semantic markup and clean TypeScript
12.5
Design— Visual aesthetics, contrast, layout harmony
12.3
Creativity— Original micro-interactions and layout depth
12.0
Fidelity— Strict adherence to complex prompt constraints
12.2