VisReason (Vision-Centric) - Inductive Reasoning: leaderboard

VisReason Inductive Reasoning is a conceptual-reasoning category of VisReason (2026): the model must infer underlying rules from observed instances and apply them to novel cases. The score is accuracy in percent, higher is better.

Metric: Accuracy (%) on the Inductive Reasoning category (conceptual reasoning: infer underlying rules from observed instances and apply them to novel cases) 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 May 2027. 20 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 3 Pro (Preview)Low53.1#119
2Qwen 3 VL 235B A22B (Thinking)46.2#459
3GPT-5.2Thinking42.3#185
4GPT-5 MiniHigh41.8#364
5GPT-5Low40.7#178
6Qwen 3 VL 32B (Thinking)38.6#573
7O4 Mini36.4#368
8Qwen 3 VL 8B (Thinking)35#910
9Qwen 3 VL 30B A3B (Thinking)34.9#719
10Qwen 3 VL 8B Instruct33.7#1029
11Gemini 2.5 Flash31.6#463
12GPT-5 Nano29.8#906
13Qwen 2.5 VL 32B Instruct25.1#1062
14Keye-VL-1.5-8BThinking24.9#1720
15MiMo-VL-7B-RL24.7#1155

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