GDPx: Measuring Frontier Model Capabilities Across the Most Valuable Domains

What happened: SemiAnalysis introduced GDPx, a benchmark covering domains such as chip design, cybersecurity, biology, finance, legal, quant trading, RSI, hardware design, materials science, alignment, and software engineering, built with 14 specialized data partners. Key numbers: Best model scores 12% on GDPx. Why it matters: The ranking aligns with subjective impressions: OpenAI and Anthropic dominate, while SpaceXAI, Meta, and open-weight models trail. Safety classifiers can lower scores, and Gemini 4 Carbon is the only model near Opus and Astra.

SemiAnalysis introduced GDPx, a benchmark covering domains such as chip design, cybersecurity, biology, finance, legal, quant trading, RSI, hardware design, materials science, alignment, and software engineering, built with 14 specialized data partners.

The ranking aligns with subjective impressions: OpenAI and Anthropic dominate, while SpaceXAI, Meta, and open-weight models trail. Safety classifiers can lower scores, and Gemini 4 Carbon is the only model near Opus and Astra.