AirGroundBench - Viewpoint Reasoning: leaderboard
Metric: Accuracy (%) on the viewpoint spatial-reasoning VQA task; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 397B A17B | 48.35 |
| 2 | Seed 2.0 Pro | 44 |
| 3 | Qwen 3.5 27B | 43.69 |
| 4 | Qwen 3.5 Plus | 41.18 |
| 5 | Claude Sonnet 4.6 | 39.08 |
| 6 | Kimi K2.5 | 38.24 |
| 7 | GPT-5.2 | 29.41 |
| 8 | GLM-4.6V | 26.73 |
| 9 | ERNIE 5.0 | 24.3 |
| 10 | Qwen 3 VL 235B A22B | 24.15 |
| 11 | Gemini 3.1 Pro (Preview) | 22.4 |
Interactive version: theaggregate.ai/benchmark?slug=airgroundbench-viewpoint-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-29.