VisReason (Vision-Centric) - Bounding-Box: leaderboard
Metric: Accuracy (%) on the 140 bounding-box questions (boxes Hungarian-matched to the ground truth at IoU above 0.5, partial credit per question) of VisReason, drawn from the Localized Reasoning and Spot the Difference categories, 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 July 2027. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) (Low) | 24.5 |
| 2 | GPT-5.2 (Thinking) | 14.7 |
| 3 | GPT-5 (Low) | 8.9 |
| 4 | GPT-5 Mini (Low) | 7.4 |
| 5 | GPT-5 Mini (High) | 6.8 |
| 6 | Qwen 3 VL 8B (Thinking) | 5.2 |
| 7 | GPT-5 Mini (Medium) | 4.8 |
| 8 | Qwen 3 VL 235B A22B (Thinking) | 4.4 |
| 9 | GPT-4o | 3.6 |
| 10 | O4 Mini | 3.3 |
| 11 | Qwen 2.5 VL 32B Instruct | 2.4 |
| 12 | GPT-5 Nano | 2 |
| 13 | Qwen 3 VL 32B (Thinking) | 1.3 |
| 14 | Qwen 3 VL 8B Instruct | 1.2 |
| 15 | InternVL3-14B | 0.7 |
Interactive version: theaggregate.ai/benchmark?slug=visreason-vision-centric-bounding-box · How It Works · Data refreshed daily, snapshot 2026-10-07.