SemiAnalysis unveils GDPx benchmark, and the best AI model scores 12%

SemiAnalysis has unveiled GDPx, a new benchmark designed to measure the general agentic capabilities of AI models, with the top-performing model scoring just 12%. The low score underscores how early AI systems remain in autonomous task execution, a key variable for crypto-AI protocols whose token valuations and total value locked (TVL) often track perceived progress in agentic AI. The announcement, covered by Crypto Briefing, arrives as AI-linked tokens and agent platforms continue to attract capital across chains, making independent benchmarks like GDPx a reference point for assessing whether agent-driven products can deliver real economic output. Market participants will watch whether subsequent model releases close the gap toward higher GDPx scores, as improvements could shift sentiment and flows toward AI-focused protocols and their tokens.