Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Anonymous Gestalt Tops Contra Frontier Design Benchmark

AI summary

Contra Labs and BenchFlow launched the Frontier Design Benchmark, covering 33 real commercial design tasks. The evaluation involved 35 professional designers, five rounds of blind testing and 4,311 votes. Contra Labs’ published results ranked the anonymous, stealth-development model Gestalt first with an Elo score of 1931 and a 90% win rate, ahead of the human reference Human golden, Claude Fable 5.2 and GPT-6.1 Sol. Gestalt did not score highest on every metric: its 83-point Design Critique Rubric score was below Claude Fable 5.2 and Claude Opus 5.5 at 84. Contra Labs has not disclosed Gestalt’s developer, architecture or identity.

In-depth insight

The benchmark compares professional designers’ preferences on real commercial design work rather than simple prompt generation, providing a direct comparison between named models, an anonymous model and a human reference.

IT之家 11 月 21 日消息,Contra Labs 于当地时间 11 月 18 日发布 AI 设计能力评测榜单 Frontier Design Benchmark。

其中,一款代号为 Gestalt、目前处于隐身开发状态(Stealth)的匿名 AI 模型成为最大黑马,在专业设计师盲测中以明显优势登顶,超过 Anthropic 的 Claude Fable 5.2、OpenAI 的 GPT-6.1 Sol 等主流模型,甚至在设计师偏好评分中超过了人类设计师的参考作品。

根据 Contra Labs 官方公布的评测结果,Gestalt 的 Elo 评分达到 1931 分,胜率为 90%,位居榜单第一。

排名第二的是人类设计师参考作品 Human golden,Elo 评分为 1739 分,胜率为 74%;Claude Fable 5.2 以 1713 分排名第三,GPT-6.1 Sol 则以 1668 分排名第五。

#Frontier Design Benchmark#Gestalt#Contra Labs#BenchFlow#Claude Fable 5.2#GPT-6.1 Sol#Human golden

Original · 2 min readRead the original
Group chat

No comments yet. Start the conversation.