No AI agent scores above 12 percent on SemiAnalysis' new real-world work benchmark
Key point
In the new GDPx benchmark released by SemiAnalysis, AI agents performing tasks such as chip design and quant trading all scored below 12 percent.
SemiAnalysis has introduced a new benchmark called GDPx, which puts AI agents through simulated knowledge work covering areas from chip design to quant trading. In this benchmark, no model broke the 12 percent mark.
Claude Opus 5.5 narrowly led over Gemini 4 Carbon. However, critics have pointed to the small number of tasks per domain and uneven testing conditions.
This item is a summary. The full, checkable version is at the original source:
Read the original at The Decoder ↗