BioBasicsBench
by OpenAI with Ligo
What is BioBasicsBench? BioBasicsBench is an agents & tool use eval made by OpenAI with Ligo. A wet-lab analysis benchmark of 100 tasks built with Ligo: agents interpret raw experimental evidence such as gels, colony plates, microscopy and instrument readouts, and are graded against expert rubrics that give partial credit for checking controls and recognising when the evidence cannot support a conclusion. OpenAI's 6 Nov 2026 post reports seven of its models across 37 settings (18,500 graded answers); at the xhigh effort setting, GPT-6 Astra earned a mean 66.7% of available credit versus 43.8% for GPT-5.6 Sol. The tasks are not released for outside runs, so every published score is OpenAI's own.
Editor note
A wet-lab analysis benchmark of 100 tasks built with Ligo: agents interpret raw experimental evidence such as gels, colony plates, microscopy and instrument readouts, and are graded against expert rubrics that give partial credit for checking controls and recognising when the evidence cannot support a conclusion. OpenAI's 6 Nov 2026 post reports seven of its models across 37 settings (18,500 graded answers); at the xhigh effort setting, GPT-6 Astra earned a mean 66.7% of available credit versus 43.8% for GPT-5.6 Sol. The tasks are not released for outside runs, so every published score is OpenAI's own.
- What it tests
- Whether AI agents can read routine wet-lab evidence, spot control problems and avoid unsupported conclusions.
- Original source
- openai.com/index/measuring-how-ai-agents-handle-wet-lab-science/
- Last updated
- Published 2026-11-06
- License
- Not checked yet — see source
Related agents & tool use evals
- GAIA — General assistant skills: browsing, file handling, multi-step reasoning.
- WebArena — Autonomous web navigation and task completion.
- OSWorld — Computer-use agents operating real operating systems and apps.
- Terminal-Bench — Agents working in the shell end to end.
Discussion
No one has discussed BioBasicsBench yet. Have you run it? Share what it caught, what it missed or how you set it up.