OpenAI推出湿实验室科学基准 BioBasicsBench

Our mission is to ensure that artificial general intelligence benefits all of humanity. As part of that mission, we want to help scientists accelerate biological research—from understanding disease to developing new treatments. For AI agents to contribute, they need to reliably interpret the experimental evidence scientists work with every day in wet labs, where they run experiments with biological samples. Everyday wet-lab work involves failed experiments, ambiguous images and conflicting observations. Existing benchmarks are built around charts and images selected for scientific papers may miss this ambiguity and the troubleshooting it requires. That’s why we’re introducing BioBasicsBench, a new benchmark developed with Ligo, a biotech company focused on enzyme design, to measure how [...] The benchmark’s 100 tasks test whether agents can interpret raw experimental data, identify errors, and recognize when the evidence cannot support a conclusion. Getting these tasks right matters be

# Measuring how AI agents handle everyday wet-lab science openai.com9 min read OpenAI and biotech company Ligo have introduced BioBasicsBench, a new benchmark designed to measure how well AI agents interpret raw experimental data across 100 routine wet-lab tasks. Evaluations of seven models showed that GPT-6 Astra achieved the highest mean reward of 66.7% at its highest reasoning effort, demonstrating improvements in multimodal interpretation and control recognition, though challenges with over-interpretation and accurate counting remain.Open article → ## More on this topic [...] ## More on this topic GPT-6 Astra caught a wrong antibiotic label on a DNA-cloning control plate that GPT-5.6 Sol misse...x.comNov 6, 20261 min A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench....x.comNov 6, 20261 min

A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench. Developed with Ligo Biosciences, a biotech ...Read more

# OpenAI’s Post View organization page for OpenAI 12,146,829 followers Report this post A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench. Developed with Ligo Biosciences, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify errors, and recognize when the evidence doesn’t support a conclusion. Accelerating biological research starts with getting the everyday work right. BioBasicsBench shows where models need to improve, and helps us track what those gains mean for scientific progress, biosafety, and biosecurity. Measuring how AI agents handle everyday wet-lab science [...] Andrew Austin 36m Report this comment Finally a test that actually matters. Can it spot the error and say nope, the data does not back that up. That is the hard part, not sounding smart. Like Reply 1 Reaction Aura Nebulai 2h Report this comment

## Conversation @OpenAI OpenAI@OpenAI A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench. Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify errors, and recognize when the evidence doesn’t support a conclusion. Accelerating biological research starts with getting the everyday work right. BioBasicsBench shows where models need to improve, and helps us track what those gains mean for scientific progress, biosafety, and biosecurity. openai.com/index/measurin… 7:08 PM · Nov 6, 2026·181.2K Views 453 @OpenAI OpenAI@OpenAI [...] Views 453 @OpenAI OpenAI@OpenAI A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench. Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can

@OpenAI OpenAI@OpenAI 3h A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench. Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify 7:54 PM · Nov 6, 2026·1,879 Views 8 @rohanpaul_ai Rohan Paul@rohanpaul\_ai [...] @OpenAI OpenAI@OpenAI 3h A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench. Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify 7:54 PM · Nov 6, 2026·1,879 Views 8 @rohanpaul_ai Rohan Paul@rohanpaul\_ai 2h BioBasicsBench artwork with a microscope and laboratory plate over a green background. Measuring how AI agents handle everyday wet-lab scienceFrom openai.com @leononrails NotaDEV \ fire