OmniStarPro-COQ (Online): leaderboard
Contextual online question answering on OmniStarPro (Yang et al., 2026): real-time questions during a live stream that depend on recent history, combining the current frame with the preceding context. A GPT-4o judge scores semantic correctness on a 0-10 scale; higher is better.
Metric: Semantic correctness (0-10; GPT-4o judge score of real-time answers to contextual questions that depend on recent stream history). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | LiveStarPro | 5.97 | |
| 2 | LiveStar | 5.85 | #1076 |
| 3 | MMDuet | 5.71 | |
| 4 | VideoLLM-online | 4.01 | |
| 5 | VideoLLM-MoD | 3.99 |
Interactive version: theaggregate.ai/benchmark?slug=omnistarpro-coq-online · How It Works · Data refreshed daily, snapshot 2026-11-12.