VLA Harness CALVIN (ABCD→D): leaderboard
Metric: Avg Len (subtasks). Source: allenai.github.io. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | MDT | 4.52 |
| 2 | RoboVLMs | 4.49 |
| 3 | UP-VLA | 4.42 |
| 4 | ReconVLA | 4.23 |
| 5 | GR-1 | 4.21 |
| 6 | Robo-Flamingo | 4.09 |
| 7 | RT-1 | 2.45 |
| 8 | MCIL | 0.4 |
Interactive version: theaggregate.ai/benchmark?slug=vla-harness-calvin-abcd-d · How It Works · Data refreshed daily, snapshot 2026-10-09.