OpenAI tests biology judgment, and GPT-5.6 clears only 31.5%
GeneBench-Pro uses messy biological datasets to test whether models can make research decisions, not just answer textbook questions.
Benchmarks News and Analysis
1 verified stories covering Benchmarks, product updates and industry developments.