Vinoground (Video Score): leaderboard

Vinoground temporal counterfactual benchmark, video score measuring whether VLMs pick the correct video for a given caption.

Metric: Video Score. Source: huggingface.co. Saturation forecast: Estimated already saturated. 24 models tracked.

Top models

#ModelScoreOverall rank
1GPT-4o38.2#343
2Qwen 2 VL 72B32.6#460
3Qwen 2 VL 7B32.4#750
4Claude 3.5 Sonnet28.8#350
5Gemini 1.5 Pro22.6#445

No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.

Interactive version: theaggregate.ai/benchmark?slug=vinoground-video-score · How It Works · Data refreshed daily, snapshot 2026-10-20.