Hunyuan-Vision: benchmark results
Provider: Tencent.
Aggregate Rating 1466 ± 52, rank #3128 of 10489 rated models, from 26 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CoreCognition Image + Video Boundary | 79.12 | Accuracy (%) | 68.9 |
| CoreCognition Image + Video Perceptualconstancy | 68.22 | Accuracy (%) | 65.5 |
| CoreCognition Image + Video Continuity | 60.76 | Accuracy (%) | 65.1 |
| CoreCognition Image + Video Intuitivephysics | 54.39 | Accuracy (%) | 54.2 |
| CoreCognition Image + Video Toolusing | 55.92 | Accuracy (%) | 52.1 |
| CoreCognition Image + Video Mechanicalreasoning | 53.67 | Accuracy (%) | 42.9 |
| CoreCognition Image + Video Intentionality | 69.39 | Accuracy (%) | 42 |
| AirGroundBench - Scene Perception | 89.66 | Accuracy (%) on the scene-perception VQA task; higher is bet | 37.5 |
| AirGroundBench - Viewpoint Reasoning | 26.13 | Accuracy (%) on the viewpoint spatial-reasoning VQA task; hi | 33.3 |
| CoreCognition Image + Video Mean | 49.76 | Accuracy (%) | 31.9 |
| CoreCognition Image + Video Hierarchy | 51.78 | Accuracy (%) | 31.1 |
| CoreCognition Image + Video Conservation | 36.94 | Accuracy (%) | 29 |
Interactive version: theaggregate.ai/model?slug=hunyuan-vision · How It Works · Data refreshed daily, snapshot 2026-11-20.