Two unnamed AI models appear on Frontier Design Benchmark, outrank Fable 5.2
Two unidentified 'stealth' AI models have appeared on the Frontier Design Benchmark leaderboard, scoring ahead of Fable 5.2. The benchmark evaluates models on real-world, high-value design projects, with scoring done by professional designers from the Contra Labs network using a design critique rubric and a catalogue of AI slop patterns.
GoKawiil's interpretation of the reporting above, not reported fact.
The appearance of unnamed models ahead of an established competitor suggests labs may be testing new systems before public release, a common practice to gauge performance without revealing branding. The benchmark's creators report a 0.94 correlation between their rubric scores and designer ratings, which could lend credibility to design-specific AI evaluation as an alternative to generic preference tests.
- Two unidentified AI models now outscore Fable 5.2 on the Frontier Design Benchmark leaderboard.
- The benchmark uses real design projects judged by professional designers from Contra Labs.
- A design critique rubric shows 0.94 correlation with designer ratings, aiming to add explainability beyond pairwise comparisons.
Source: contralabs.com, 2026-11-19
Published there as: “Two stealth models just appeared on the leaderboard, both ahead of Fable 5.2”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.