Model evaluation results for System One Mosaic Benchmark (S1MB), combining specialized Choice, Noul, and Score tasks. Each / folder represents one leaderboard entry and contains metadata.json plus per-benchmark.json.xz files. Results retain their evaluation settings and model, dataset, and evaluator provenance. To add or update results, follow the for evaluation, export, validation, and opening a pull request to this dataset. See the S1MB project for code and documentation.