Research
SemiAnalysis launches GDPx benchmark where no AI agent scores above 12 percent
The GDPx benchmark tests agents on simulated knowledge work from chip design to quant trading. Claude Opus 5.5 narrowly leads Gemini 4 Carbon, and critics question the small task count per domain and uneven testing conditions.
Every outlet that ran it 1 article
The DecoderTech pressLead
No AI agent scores above 12 percent on SemiAnalysis' new real-world work benchmark