CrossVid - Temporal Understanding: leaderboard
Mean accuracy over CrossVid's three temporal tasks: inferring plot, aligning functional steps and ordering procedural steps across videos.
Metric: Accuracy (%). Source: raw.githubusercontent.com. Saturation forecast: Estimated already saturated. 16 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 56 | #274 |
| 2 | GPT-4.1 | 46.7 | #505 |
| 3 | Video-R1-7B | 26.9 | #2053 |
| 4 | Qwen 2.5 VL 32B Instruct | 26.5 | #1074 |
| 5 | MiniCPM-o-2.6 | 26.4 | #2417 |
| 6 | GLM-4.1V-9B (Thinking) | 23.1 | |
| 7 | LLaVA-Video-72B-Qwen2 | 22 | #1344 |
| 8 | Qwen2.5-Omni-7B | 21.6 | #1883 |
| 9 | InternVL3-8B | 20.3 | #1756 |
| 10 | Qwen 2.5 VL 7B Instruct | 20 | #1893 |
| 11 | Kimi-VL-A3B (Thinking) | 17.9 | |
| 12 | Phi-3.5-Vision Instruct | 17.2 | #6271 |
| 13 | InternVL3-78B | 15.6 | #755 |
| 14 | InternVL3-38B | 10.1 | #927 |
| 15 | LongVA-7B-DPO | 7.5 | #2177 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=crossvid-temporal-understanding · How It Works · Data refreshed daily, snapshot 2026-11-17.