MARINER - VQA Environmental State Inference: leaderboard
Maritime visual question answering on MARINER (2026), with multiple-choice questions built from verified image metadata and audited by maritime annotators. Accuracy in percent on environmental state questions; higher is better.
Metric: Accuracy (%) on MARINER maritime visual question answering (environmental state 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. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 92.38 | #276 |
| 2 | InternVL3-78B | 92.31 | #759 |
| 3 | InternVL3-38B | 91.82 | #939 |
| 4 | LLaVA-OneVision-72B | 91.82 | #1771 |
| 5 | Qwen 2.5 VL 7B Instruct | 91.33 | #1902 |
| 6 | Qwen 2.5 VL 72B Instruct | 91.26 | #841 |
| 7 | InternVL3-8B | 90.85 | #1782 |
| 8 | MiniCPM-V-2.6 | 90.78 | #3267 |
| 9 | Gemini 2.5 Flash | 90.08 | #463 |
| 10 | GPT-4.1 | 90.01 | #505 |
| 11 | Qwen 2.5 VL 3B Instruct | 89.59 | #2713 |
| 12 | LLaVA-OneVision-7B | 89.03 | #2981 |
| 13 | InternVL2-8B | 88.61 | #3227 |
| 14 | GPT-4o | 85.32 | #751 |
| 15 | LLaVA-v1.5-7B | 79.25 | #7630 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=mariner-vqa-environmental-state-inference · How It Works · Data refreshed daily, snapshot 2026-11-21.