SurgVLA-Bench - Level 1 Electrocoagulation: leaderboard
Contact-based cauterization of a target tissue site with a laparoscopic instrument in the SurRoL simulator, an atomic task of SurgVLA-Bench (Sun et al., 2026). Vision-language-action models are LoRA fine-tuned on the Level 1 demonstrations; binary success rate in percent over 50 trials with a 100-step limit; higher is better.
Metric: Success rate (%; T1-3 electrocoagulation of a target tissue site, SurgVLA-Bench Level 1, 50 trials in the SurRoL laparoscopic simulator with a 100-step limit; LoRA fine-tuned on the level's demonstrations). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | OpenVLA | 76 | |
| 2 | π0 | 0 | |
| 3 | π0.5 | 0 | |
| 4 | SmolVLA | 0 |
Interactive version: theaggregate.ai/benchmark?slug=surgvla-bench-level-1-electrocoagulation · How It Works · Data refreshed daily, snapshot 2026-11-12.