Measuring how AI agents handle everyday wet-lab science
benchmarksOpenAI NewsNov 6, 2026
OpenAI introduces BioBasicsBench, a 100-task benchmark measuring how well AI agents interpret raw experimental data and handle everyday wet-lab science.
Impact: LowConfidence: MediumNovelty: Incremental
benchmarksopenaiagents