OVO-S-Bench: leaderboard
Metric: Overall 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), macro-average of the four levels. Source: arxiv.org. Saturation forecast: Around 2031. 31 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 59.19 |
| 2 | Qwen 3 VL 235B A22B | 53.64 |
| 3 | Qwen 3.5 397B A17B | 52.1 |
| 4 | Qwen 3.5 27B (Non-reasoning) | 51.69 |
| 5 | GPT-5.4 | 50.89 |
| 6 | Gemini 3.1 Flash Lite | 50.81 |
| 7 | Qwen 3 VL 32B Instruct | 48.8 |
| 8 | Qwen 3 VL 4B Instruct | 46.83 |
| 9 | Qwen 3.5 9B (Non-reasoning) | 45.91 |
| 10 | Qwen 3.5 4B (Non-reasoning) | 45.4 |
| 11 | InternVL3.5-8B | 44.57 |
| 12 | Qwen 2.5 VL 7B Instruct | 44.19 |
| 13 | Grok 4.1 Fast (Non-reasoning) | 43.73 |
| 14 | Gemma 4 26B A4B | 42.56 |
| 15 | Gemma 4 E4B | 39.69 |
Interactive version: theaggregate.ai/benchmark?slug=ovo-s-bench · How It Works · Data refreshed daily, snapshot 2026-09-29.