OpenAI's latest benchmark, GDPval, suggests its GPT-5 and Anthropic's Claude Opus 4.1 are nearing human expert performance in various sectors. GPT-5 high achieved a 35.48% win rate, while Claude Opus 4.1 excelled with 43.56%. This test evaluated models across 44 occupations in industries contributing significantly to US GDP. OpenAI emphasizes that while AI can perform certain tasks efficiently, many jobs require more than simple task execution, indicating a complementary future for AIs and human workers.
OpenAI's latest benchmark, GDPval, suggests its GPT-5 and Anthropic's Claude Opus 4.1 are nearing human expert performance in various sectors. GPT-5 high achieved a 35.48% win rate, while Claude Opus 4.1 excelled with 43.56%. This test evaluated models across 44 occupations in industries contributing significantly to US GDP. OpenAI emphasizes that while AI can perform certain tasks efficiently, many jobs require more than simple task execution, indicating a complementary future for AIs and human workers.