V-STaR: leaderboard
Spatio-temporal reasoning benchmark for Video-LLMs, reporting category-level modified arithmetic mean (mAM) and modified logarithmic geometric mean (mLGM) over video question chains.
Metric: Average All Score. Source: huggingface.co. Saturation forecast: Around February 2027. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 32.45 | #741 |
| 2 | Gemini 2.0 Flash | 31.22 | #702 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=v-s-tar · How It Works · Data refreshed daily, snapshot 2026-11-12.