MARINER - VQA Attributes: leaderboard
Metric: Accuracy (%) on MARINER maritime visual question answering (vessel attribute 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-4.1 | 87.46 |
| 2 | Gemini 2.5 Pro | 86.02 |
| 3 | GPT-4o | 82.97 |
| 4 | InternVL3-78B | 80.84 |
| 5 | Qwen 2.5 VL 72B Instruct | 80.4 |
| 6 | InternVL3-38B | 80.36 |
| 7 | Qwen 2.5 VL 32B Instruct | 75.96 |
| 8 | Gemini 2.5 Flash | 75.09 |
| 9 | Qwen 2.5 VL 7B Instruct | 74.26 |
| 10 | InternVL3-8B | 71.56 |
| 11 | MiniCPM-V-2.6 | 68.51 |
| 12 | InternVL2-8B | 63.81 |
Interactive version: theaggregate.ai/benchmark?slug=mariner-vqa-attributes · How It Works · Data refreshed daily, snapshot 2026-10-07.