CausalPhys - Intervention: leaderboard
Metric: Answer accuracy (0-1 scaled to %) on the Intervention domain (reasoning under explicit manipulations such as viewpoint, spatial or temporal changes) of the image- and video-based questions with verified answers over real-world physical scenes; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o | 57.07 |
| 2 | InternVL3-78B | 57.07 |
| 3 | Claude Sonnet 4 | 56.72 |
| 4 | Gemini 2.5 Flash | 55.67 |
| 5 | GPT-4o Mini | 51.31 |
| 6 | Mistral Small 3.2 | 48.52 |
| 7 | Phi-4 Multimodal Instruct | 48.52 |
| 8 | Qwen 3 VL 32B | 47.99 |
| 9 | Qwen 2 VL 7B | 47.64 |
Interactive version: theaggregate.ai/benchmark?slug=causalphys-intervention · How It Works · Data refreshed daily, snapshot 2026-09-29.