GraphVerse - Longest Cycle - VGR-Score: leaderboard

Metric: Visual graph reasoning score (0-100; 0.7 process + 0.3 answer, GPT-5.1 verifier). Source: arxiv.org. 19 models tracked.

Top models

#ModelScore
1GPT-5.291.8
2Gemini 3 Pro89.7
3Gemini 3 Flash79.2
4GLM-4.6V72.7
5O170.2
6Kimi K2.5 (Thinking)67.7
7Qwen 3.5 4B44.8
8QVQ-72B-Preview44.4
9Grok 4 Fast39.9
10InternVL3-8B31.7
11Qwen 3.5 35B A3B20.7

Interactive version: theaggregate.ai/benchmark?slug=graphverse-longest-cycle-vgr-score · How It Works · Data refreshed daily, snapshot 2026-09-19.