Apple-π benchmark tests whether video models can reason about physics
liuziwei7 · x · 2026-07-21
Researchers introduce Apple-π, a benchmark for asking whether video models can reason about physics instead of merely producing plausible motion.
- The benchmark is designed around law-grounded physical reasoning.
- It decomposes the task into Perception → Formulation → Deduction.
- The dataset contains 400 videos and 10 mechanics tasks.
- The authors released the paper, code, and leaderboard, making it usable for evaluation and follow-up work.
Related event: Apple-π Benchmark Tests Video Models' Physics Reasoning(3 posts)→
More from Research
- Paper: data schedules and learning rate schedules should be designed together — alifcoder · 2026-10-27
- Health benchmark study: quality and diversity of eval items beat quantity — keunwoochoi · 2026-10-27
- Across 7 tasks, highest reasoning setting isn't the best for most open models — keunwoochoi · 2026-10-27
- Open Models Match GPT-6 on MedMCQA but Trail 17 Points on Hard Clinical Cases — keunwoochoi · 2026-10-27
- Health-Focused Benchmark Reranks LLMs: Muse Spark Open Jumps from #9 to #1 — keunwoochoi · 2026-10-27
- Open-source Latch compiles PyTorch into training megakernels, 2.75x faster on H100 — sguada · 2026-10-27