WideSearch: Results as Published

WideSearch tests whether AI search agents can gather a large amount of information completely and accurately, such as a full table of facts about many entities, rather than a single answer. It has 200 questions in English and Chinese across more than 15 domains, drawn from real user queries, and each asks for many facts that are checked one at a time. The main score is an exact match to the reference table, and most systems scored close to zero while the best scored about 5 percent.

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.