Stanford releases Terminal-Bench-Science: AI agent benchmark for research workflows
BenBlaiszik · x · 2026-08-28
A Stanford-led community effort has released Terminal-Bench-Science, a benchmark designed to evaluate AI agents on real-world research workflows across scientific domains.
- Background: Built by the Terminal-Bench team in collaboration with scientific domain experts globally.
- Scale: The v0.1 release includes 70 tasks.
- Performance: Even the top-performing model, Claude Opus 5, currently solves only 30% of the tasks.
This benchmark aims to assess agents' ability to handle practical scientific research processes rather than just Q&A.
Related event: Stanford-led Terminal-Bench-Science Debuts; Claude Opus 5 Passes Just 30%(19 posts)→
More from Research
- Interpretability paper incoming: in-depth look at testing explanations on modern LLMs — burny_tech · 2026-10-16
- Better-learning models tend to be more interpretable, argues Nick Cammarata — burny_tech · 2026-10-16
- 'Interp was mostly hopeless before — now it's automatable': Cammarata on interpretability — burny_tech · 2026-10-16
- Cornell and Amazon revisit the 16x pretraining compute gap: diffusion LMs need only 3-8x — LucaAmb · 2026-10-16
- AI trained on the fraction of human intelligence that fit the page — r0ck3t23 · 2026-10-16
- Quotient Recurrent Reasoners hit 97.9% on Sudoku-Extreme with 78% fewer loops — burny_tech · 2026-10-16