Progress-Bench - Human Activities NSE: leaderboard
Metric: Normalized score error (%; pointwise error of the predicted progress score; 587 in-the-wild human-activity samples, mean of vision- and text-demonstration results; direct prediction). Source: arxiv.org. Saturation forecast: Around December 2026. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 VL 8B | 22.3 |
| 2 | Qwen 2.5 VL 7B | 29.2 |
Interactive version: theaggregate.ai/benchmark?slug=progress-bench-human-activities-nse · How It Works · Data refreshed daily, snapshot 2026-09-26.