MARINER - VQA Spatial Relationship: 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 spatial relationship questions; higher is better.

Metric: Accuracy (%) on MARINER maritime visual question answering (spatial relationship 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-4o35.26#751
2Qwen 2.5 VL 72B Instruct35.26#841
3Qwen 2.5 VL 32B Instruct35.26#1062
4Gemini 2.5 Flash33.53#463
5InternVL3-38B33.53#939
6GPT-4.132.37#505
7Gemini 2.5 Pro31.79#276
8InternVL2-8B31.79#3227
9InternVL3-78B30.64#759
10Qwen 2.5 VL 3B Instruct30.64#2713
11InternVL3-8B29.48#1782
12Qwen 2.5 VL 7B Instruct27.75#1902
13LLaVA-OneVision-72B27.75#1771
14MiniCPM-V-2.626.01#3267
15LLaVA-OneVision-7B24.86#2981

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