MMStar: Results as Published

MMStar tests multimodal understanding on items that were kept only when the image is needed. Humans selected 1,500 samples, with 250 for each of six capabilities and 18 finer axes. The six capabilities are coarse perception, fine-grained perception, instance reasoning, logical reasoning, science and technology, and mathematics. The main result is accuracy, and the paper also defines Multi-modal Gain and Multi-modal Leakage for separate analyses.

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.