MARINER - VQA Perspective Estimation: leaderboard
Metric: Accuracy (%) on MARINER maritime visual question answering (vessel perspective questions; multiple-choice questions generated from verified image metadata and audited by maritime annotators); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 | 72.73 |
| 2 | Gemini 2.5 Pro | 68.53 |
| 3 | GPT-4o | 62.7 |
| 4 | Gemini 2.5 Flash | 59.67 |
| 5 | InternVL3-78B | 56.41 |
| 6 | Qwen 2.5 VL 72B Instruct | 56.18 |
| 7 | InternVL3-38B | 55.94 |
| 8 | Qwen 2.5 VL 7B Instruct | 51.52 |
| 9 | InternVL3-8B | 50.58 |
| 10 | Qwen 2.5 VL 32B Instruct | 49.65 |
| 11 | MiniCPM-V-2.6 | 44.29 |
| 12 | InternVL2-8B | 41.72 |
Interactive version: theaggregate.ai/benchmark?slug=mariner-vqa-perspective-estimation · How It Works · Data refreshed daily, snapshot 2026-10-07.