ProtocolQA: Results as Published

ProtocolQA tests whether a model can spot and fix mistakes in published lab protocols. In the open-ended form, each question describes a protocol with an error, gives the wet-lab result of following it, and asks how the procedure should be corrected. OpenAI's GPT-5 system card describes the open-ended form as short-answer versions of 108 multiple-choice questions from FutureHouse's ProtocolQA dataset.

Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.