RoboProcessBench - Bimanual Coordination: leaderboard
Metric: Accuracy (%) on 533 questions asking how the two arms are currently coordinating, from a short ordered clip (4 options, chance 25.0); zero-shot multiple-choice VQA on held-out robot manipulation recordings from GM-100, RH20T, REASSEMBLE and AIST-Bimanual (ProcessData-Eval), temperature 0.01; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 VL 32B | 41.9 |
| 2 | GPT-4o | 40 |
| 3 | Claude Haiku 4.5 | 39.5 |
| 4 | GPT-5.4 Mini | 38.1 |
| 5 | Qwen 2.5 VL 7B Instruct | 36.4 |
| 6 | Claude Sonnet 4.6 | 36.3 |
| 7 | InternVL3.5-8B | 26.5 |
| 8 | GLM-4.6V | 26 |
| 9 | InternVL3-38B | 25.1 |
| 10 | InternVL3-8B | 22.9 |
Interactive version: theaggregate.ai/benchmark?slug=roboprocessbench-bimanual-coordination · How It Works · Data refreshed daily, snapshot 2026-09-29.