ImplicitQA (Macro Average Accuracy): leaderboard
ImplicitQA video-reasoning benchmark scored as accuracy macro-averaged across its nine implicit-reasoning dimensions (spatial, depth, viewpoint, motion, causal/motivational, social, physical context, inferred counting).
Metric: Macro Average Accuracy. Source: huggingface.co. Saturation forecast: Estimated already saturated. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 | 68.6 | #124 |
| 2 | GPT-4.1 | 58.6 | #248 |
| 3 | Qwen 2.5 VL 7B | 46.1 | #643 |
| 4 | Qwen 2 VL 7B | 46 | #750 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=implicitqa-macro-average-accuracy · How It Works · Data refreshed daily, snapshot 2026-10-20.