VisReason (Vision-Centric) - Localized Reasoning: leaderboard

VisReason Localized Reasoning is a perceptual-reasoning category of VisReason (2026): the model must infer the target instance in an image and localize it with a bounding box. The score is accuracy in percent, higher is better.

Metric: Accuracy (%) on the Localized Reasoning category (perceptual reasoning: infer the target instance in an image and localize it with a bounding box) 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)Low44.2#123
2GPT-5.2Thinking24.6#190
3GPT-5Low15#185
4GPT-5 MiniHigh12.5#373
5Qwen 3 VL 8B (Thinking)9.6#924
6Qwen 3 VL 235B A22B (Thinking)7.1#462
7GPT-4o6.3#754
8GPT-5 Nano3.8#912
9O4 Mini3.8#372
10Qwen 2.5 VL 32B Instruct3.3#1065
11Qwen 3 VL 2B (Thinking)2.5#2552
12MiMo-VL-7B-RL2.1#1160
13Qwen 3 VL 8B Instruct1.3#1043
14InternVL3-14B1.3#1227
15Qwen 3 VL 32B (Thinking)1.3#577

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=visreason-vision-centric-localized-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-22.