OpenAI introduces BioBasicsBench, a 100-task test of AI lab analysis

On Nov. 6, 2026, OpenAI introduced BioBasicsBench, a benchmark designed to evaluate how AI agents handle routine analysis in wet-lab science.

In its announcement, OpenAI described a test developed with Ligo, a biotech company focused on enzyme design. The benchmark draws on 100 tasks from real experiments.

Those tasks test whether agents can interpret raw lab data, identify errors and recognize when the evidence does not support a conclusion, according to OpenAI. That last target makes recognizing insufficient evidence part of the evaluation, alongside interpreting the data and spotting mistakes.

OpenAI framed the benchmark as a way to identify where models need improvement and track what those gains mean for scientific progress, biosafety and biosecurity. Its announcement gives no model scores or measured research or safety outcomes.