VisReason (Vision-Centric) - 3D-Spatial Reasoning: leaderboard

VisReason 3D-Spatial Reasoning is a structural-reasoning category of VisReason (2026): the model must reason about implicit 3D geometric relationships to infer the correct answer. The score is accuracy in percent, higher is better.

Metric: Accuracy (%) on the 3D-Spatial Reasoning category (structural reasoning: reason about implicit 3D geometric relationships to infer the correct answer) 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 2030. 20 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3 Pro (Preview)Low43.9#119
2Qwen 3 VL 235B A22B (Thinking)36.2#459
3Qwen 3 VL 8B (Thinking)33.1#910
4Gemini 2.5 Flash31.5#463
5InternVL3-8B31.5#1782
6Qwen 3 VL 30B A3B (Thinking)30.8#719
7Qwen 3 VL 32B (Thinking)30.8#573
8GPT-5.2Thinking30.8#185
9GPT-5 MiniLow30.8#364
10O4 Mini27.7#368
11Qwen 2.5 VL 32B Instruct27.7#1062
12GPT-5Low26.2#178
13GPT-4o25.4#751
14MiMo-VL-7B-RL25.4#1155
15GPT-5 Nano23.9#906

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-3d-spatial-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-21.