RoboMME (VideoQA): leaderboard
Metric: Success rate (%) averaged over RoboMME's 16 memory-dependent manipulation tasks (counting, object permanence, reference and imitation) reformulated as online video question answering: the model watches the execution video revealed step by step and picks the next high-level action, which an oracle planner executes; 50 episodes per task, one run each; higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 47.88 | #145 |
| 2 | GPT-4o | 30.63 | #333 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=robomme-videoqa · How It Works · Data refreshed daily, snapshot 2026-10-11.