V* Bench: Results as Published

V* Bench tests whether a multimodal model can find small visual details in high-resolution images. It has 191 images, with an average resolution of 2246 by 1582, and two sub-tasks: attribute recognition on 115 samples and spatial-relationship reasoning on 76 samples. Attribute questions ask about a property of an object, such as color or material, and spatial questions ask how two objects are placed relative to each other. The project page presents the questions as multiple choice, with one option marked correct.

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.