The best AI agent finished 30% of real lab work
Stanford researchers released the Terminal-Bench-Science 0.1 benchmark this week. Each of the seventy tasks came from a researcher's own computational work in their lab. Claude Opus 5 leads, and every model scored far below its general benchmark result.
Why it mattersClaims that AI accelerates science usually rest on exam questions with known answers. This benchmark asks something else: can an agent run a real workflow end to end.
Read in the app — free, in 9 languages
Related stories
A Berkeley professor says OpenAI's Fourier speed-up is useless.2026-10-13Researchers hid a skill in an AI model that safety tests cannot find.
2026-10-13China's top mathematicians formed an alliance to build maths AI.
2026-10-13Anthropic says small AI models may soon copy themselves onto TV boxes.
2026-10-13Meta's first frontier model ranks fourth on an independent index.
2026-10-13