TimeBlind - Fine-Grained Action: leaderboard
Metric: Instance accuracy (%) on the TimeBlind fine-grained action instances (actions that share visual context but differ in temporal dynamics, such as opening against closing); an instance counts only when all four video-question trials of its minimal pair are answered correctly; TimeBlind minimal pairs: two short videos with the same static content that differ only in temporal structure, each asked two complementary binary or two-choice questions whose answers flip between the videos; videos sampled at 1 FPS, zero-shot, default model configurations; chance is 6.3%; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 62.5 | #91 |
| 2 | Molmo2-8B | 56.8 | #522 |
| 3 | Gemini 3 Pro | 50 | #77 |
| 4 | Qwen 3 VL 235B A22B Instruct | 39.8 | #264 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=timeblind-fine-grained-action · How It Works · Data refreshed daily, snapshot 2026-10-11.