Gestalt tops design leaderboard with 1931 Elo
A stealth AI model codenamed Gestalt has surged to the top of Contra Labs' Frontier Design Benchmark with an unprecedented 1,931 Elo rating. Evaluated blindly by professional designers on production UI briefs, the unannounced model outperformed Claude Fable 5.2 by 218 points and cleared the human baseline.
An unannounced model beating human designers and Claude on complex design briefs signals that UI generation is nearing its own frontier benchmark inflection point.
- –Gestalt reached a 1,931 Elo rating, establishing a 218-point lead over Claude Fable 5.2 (1,713 Elo) and surpassing Contra's human professional baseline (1,739 Elo).
- –Contra Labs' Frontier Design Benchmark evaluates models using 33 multi-constraint client briefs scored in blind pairwise votes by practicing designers, making the score difficult to inflate with shallow templates.
- –The stealth release pattern mirrors arena-style blind deployments commonly used by leading labs to stress-test frontier reasoning and multimodal models prior to launch.
- –If replicated in production workflows, frontend agents may quickly move past boilerplate code generation into nuanced layout systems, visual hierarchy, and end-to-end interface design.
DISCOVERED
52m ago
2026-11-19
PUBLISHED
1h ago
2026-11-19
RELEVANCE
AUTHOR
0x0SojalSec