MARINER - VQA Attributes: 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 vessel attribute questions; higher is better.

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. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-4.187.46#505
2Gemini 2.5 Pro86.02#276
3GPT-4o82.97#751
4InternVL3-78B80.84#759
5Qwen 2.5 VL 72B Instruct80.4#841
6InternVL3-38B80.36#939
7LLaVA-OneVision-72B77.44#1771
8Qwen 2.5 VL 32B Instruct75.96#1062
9Gemini 2.5 Flash75.09#463
10Qwen 2.5 VL 7B Instruct74.26#1902
11LLaVA-OneVision-7B72.52#2981
12InternVL3-8B71.56#1782
13Qwen 2.5 VL 3B Instruct69.21#2713
14MiniCPM-V-2.668.51#3267
15InternVL2-8B63.81#3227

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-attributes · How It Works · Data refreshed daily, snapshot 2026-11-21.