BlogLLM Benchmark Scores as Published, Side by Side
IntPhys 2: Results as Published
Every result the catalog holds for IntPhys 2: 3 models from 3 developers, read from 1 page. Each number is as printed by the company named beside it; nothing is ranked or rescored. Models are listed by developer, newest first.
Area Multimodal UnderstandingMaintainer Meta FAIRIntroduced 2025Data version 2026103001, as of 2026-10-30
IntPhys 2 tests intuitive physics in video: the model watches short clips and has to tell physically possible events from impossible ones, across principles such as object permanence, immutability, spatio-temporal continuity and solidity. The score is accuracy.
Not checked against a maintainer's page.
IntPhys 2
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| Muse Spark OpenMeta, released 2026-10-15 | 68.3% | Meta, Meta Muse Spark Open 816B-A42B model cardchecked 2026-10-19 | Effort: xhigh |
| Kimi K3Moonshot AI, released 2026-07-16 | 60.3% | Meta, Meta Muse Spark Open 816B-A42B model cardchecked 2026-10-19 | Effort: max |
| InklingThinking Machines, released 2026-07-14 | 55.2% | Meta, Meta Muse Spark Open 816B-A42B model cardchecked 2026-10-19 | Effort: xhigh |
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.