VisReason (Vision-Centric) - Geolocation: leaderboard

VisReason Geolocation is a conceptual-reasoning category of VisReason (2026): the model must extract geographic cues from the visual scene to infer its location. The score is accuracy in percent, higher is better.

Metric: Accuracy (%) on the Geolocation category (conceptual reasoning: extract geographic cues from the visual scene to infer its location) of VisReason, answers scored by format (regular-expression match, GPT-5-mini judge for open-ended answers, IoU above 0.5 for boxes), zero-shot with chain-of-thought prompt templates, one per answer format; rows not shaded gray in the paper use explicit reasoning (GPT-5 and Gemini 3 Pro at low reasoning effort); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 20 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3 Pro (Preview)Low76#119
2GPT-5Low66.5#178
3GPT-5 MiniHigh61#364
4O4 Mini55.5#368
5GPT-5.2Thinking52.5#185
6Qwen 3 VL 235B A22B (Thinking)46#459
7MiMo-VL-7B-RL37.5#1155
8Qwen 3 VL 32B (Thinking)31#573
9Qwen 3 VL 8B Instruct29.5#1029
10GPT-5 Nano29#906
11Qwen 2.5 VL 32B Instruct29#1062
12Qwen 3 VL 8B (Thinking)25.3#910
13Qwen 3 VL 30B A3B (Thinking)22.5#719
14InternVL3-14B14#1214
15Gemini 2.5 Flash12.5#463

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=visreason-vision-centric-geolocation · How It Works · Data refreshed daily, snapshot 2026-11-21.