High impact

Measuring how AI agents handle everyday wet-lab science

What happened

OpenAI introduced BioBasicsBench, a new benchmark created with biotech firm Ligo to evaluate how AI agents interpret raw experimental data across 100 routine wet‑lab tasks. The tasks cover protein methods, imaging and cytometry, quantification and assays, nucleic‑acid methods, and molecular cloning, using real evidence such as gel images, colony plates, microscopy stacks and instrument readouts. Each task was authored by scientists, reviewed internally and externally, and includes expert grading criteria. At the ‘xhigh’ setting, GPT‑6 Astra achieved 66.7% of the possible credit, while GPT‑5.6 Sol reached 43.8%. The benchmark scores agents on skills like data interpretation, error detection and protocol understanding.

Why it matters

This matters because it shows AI’s growing ability to help scientists avoid costly mistakes by correctly reading lab data. As the technology improves, everyday researchers could rely on AI assistants to speed up experiments and focus on discovery.

More from OpenAI

Introducing ChatGPT WalletOpenAI · 13 Nov 2026Shield helps technicians hand off legacy portal work with OpenAIOpenAI · 13 Nov 2026OpenAI launches ChatGPT Wallet, AI agents pay for purchaseOpenAI · 13 Nov 2026OpenAI disbands its robotics team as its general models start running robotsOpenAI · 13 Nov 2026Yang-Mills, OpenAI proof, and the mass gap problemOpenAI · 13 Nov 2026

This page was collected from openai.com and summarised automatically by AIPulsee's pipeline. See how it runs.