OVO-S-Bench - Egocentric Perception: leaderboard
Metric: Level 1 (instantaneous egocentric perception) multiple-choice accuracy (%) on the 1,680-question evaluation set, prefix-access mode (128 frames sampled uniformly from the video before the query timestamp, question given with the frames), mean of its three task families. Source: arxiv.org. Saturation forecast: Around 2031. 31 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 61.92 |
| 2 | GPT-5.4 | 54.64 |
| 3 | Gemini 3.1 Flash Lite | 54.09 |
| 4 | Qwen 3 VL 235B A22B | 52.45 |
| 5 | Qwen 3.5 27B (Non-reasoning) | 51.5 |
| 6 | Qwen 3 VL 32B Instruct | 50.14 |
| 7 | Qwen 3.5 397B A17B | 49.57 |
| 8 | Gemma 4 26B A4B | 49.33 |
| 9 | Qwen 3.5 9B (Non-reasoning) | 47.47 |
| 10 | InternVL3.5-8B | 45.92 |
| 11 | Qwen 3.5 4B (Non-reasoning) | 45.41 |
| 12 | Grok 4.1 Fast (Non-reasoning) | 44.8 |
| 13 | Qwen 3 VL 4B Instruct | 43.25 |
| 14 | Gemma 4 E4B | 40.92 |
| 15 | Qwen 2.5 VL 7B Instruct | 40.67 |
Interactive version: theaggregate.ai/benchmark?slug=ovo-s-bench-egocentric-perception · How It Works · Data refreshed daily, snapshot 2026-09-29.