EmbSpatial-Bench: Results as Published

EmbSpatial-Bench tests spatial understanding from the camera's own point of view, rather than from an object drawn inside the picture. The questions are multiple choice and are built from indoor embodied scenes in MP3D, AI2-THOR, and ScanNet, then checked by people. The paper names six relations: above, below, left, right, far, and close. The reported result is accuracy, taken either from the generated choice or from the choice the model assigns the highest probability.

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.