New GDPx Test Finds No AI Agent Above 12 Percent
SemiAnalysis has released GDPx, a new test that puts AI agents through simulated knowledge work such as chip design and quant trading. No model scores above 12 percent. Claude Opus 5.5 leads narrowly over Gemini 4 Carbon, though critics say it has few tasks.
Quick glossarySemiAnalysis, benchmark, quant trading
- SemiAnalysis
- A research firm known for detailed analysis of computer chips and the AI hardware industry.
- benchmark
- A standard test used to measure and compare how well AI systems perform specific tasks.
- quant trading
- Trading that uses complex mathematical models to decide which financial assets to buy and sell.
Burning Potato score: 72%