Announcement
SemiAnalysis announces GDPx benchmark for AI-agent capabilities
SemiAnalysis puts the top model score at 12% and warns that safety classifiers can depress some results.
TLDR
SemiAnalysis introduced GDPx to measure AI-agent capabilities and put the top model score at 12%. It argues that safety classifiers depress overall scores for Opus 5.5 and GPT-6 Astra in some domains. Sundar Pichai praised the benchmark’s headroom and pointed to Carbon’s cost line, but gave no figure.
Combined views
1.7M
8 Sources, first seen ago
