ProfBench: Results as Published

ProfBench asks a model to write a long report for a professional task given in text. The paper counts 7,347 human-written response-criterion pairs across 80 tasks, split evenly across Physics PhD, Chemistry PhD, Finance MBA, and Consulting MBA. Experts with a PhD, an MBA, or equivalent experience wrote the prompts and the criteria. An LLM judge marks each criterion as fulfilled or not, and the benchmark reports performance from those judgments.

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.