A3Bench - Detailed Questions: leaderboard
Zero-shot multiple-choice video QA accuracy (%) on fine-grained detailed compositional questions in A3Bench (arXiv:2611.24762).
Metric: Detailed Questions Accuracy (%). Source: arxiv.org. Saturation forecast: Around March 2028. 14 models tracked. Updated 20 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 8B Instruct | 35.19 | #1035 |
| 2 | GPT-5.1 | 33.32 | #242 |
| 3 | Ovis2.5-9B | 33.09 | #1189 |
| 4 | Gemma 4 26B A4B (IT) | 30.86 | #561 |
| 5 | Qwen 3 VL 8B (Thinking) | 30.77 | #920 |
| 6 | InternVL3.5-8B | 28.75 | #1599 |
| 7 | InternVL3-9B | 28.22 | #1731 |
| 8 | InternVL2.5-8B | 26.75 | #2370 |
| 9 | Qwen 2.5 VL 7B Instruct | 23.73 | #1904 |
| 10 | InternVideo2.5-8B | 21.51 | |
| 11 | VideoLLaMA3-7B | 21.41 | #2000 |
| 12 | X-CLIP | 20.07 | |
| 13 | ViCLIP | 16.16 | |
| 14 | LLaVA-Video-7B-Qwen2 | 13.11 | #1977 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=a3bench-detailed-questions · How It Works · Data refreshed daily, snapshot 2026-11-20.