MARINER - VQA Operational Status: leaderboard
Metric: Accuracy (%) on MARINER maritime visual question answering (operational status 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 | 86.45 |
| 2 | InternVL3-38B | 85.35 |
| 3 | Gemini 2.5 Flash | 85.13 |
| 4 | InternVL3-78B | 83.81 |
| 5 | Qwen 2.5 VL 72B Instruct | 83.81 |
| 6 | Gemini 2.5 Pro | 83.48 |
| 7 | InternVL3-8B | 83.26 |
| 8 | Qwen 2.5 VL 32B Instruct | 82.82 |
| 9 | Qwen 2.5 VL 7B Instruct | 82.49 |
| 10 | InternVL2-8B | 80.4 |
| 11 | MiniCPM-V-2.6 | 80.18 |
| 12 | GPT-4o | 77.97 |
Interactive version: theaggregate.ai/benchmark?slug=mariner-vqa-operational-status · How It Works · Data refreshed daily, snapshot 2026-10-07.