MARINER - VQA Spatial Relationship: leaderboard
Metric: Accuracy (%) on MARINER maritime visual question answering (spatial relationship questions; multiple-choice questions generated from verified image metadata and audited by maritime annotators); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o | 35.26 |
| 2 | Qwen 2.5 VL 72B Instruct | 35.26 |
| 3 | Qwen 2.5 VL 32B Instruct | 35.26 |
| 4 | Gemini 2.5 Flash | 33.53 |
| 5 | InternVL3-38B | 33.53 |
| 6 | GPT-4.1 | 32.37 |
| 7 | Gemini 2.5 Pro | 31.79 |
| 8 | InternVL2-8B | 31.79 |
| 9 | InternVL3-78B | 30.64 |
| 10 | InternVL3-8B | 29.48 |
| 11 | Qwen 2.5 VL 7B Instruct | 27.75 |
| 12 | MiniCPM-V-2.6 | 26.01 |
Interactive version: theaggregate.ai/benchmark?slug=mariner-vqa-spatial-relationship · How It Works · Data refreshed daily, snapshot 2026-10-07.