GraphVerse - Graph Edit Distance (Paired Images) - Accuracy: leaderboard
GraphVerse (arXiv 2608.06769v1, Table 1): multimodal models solve for the edit distance between two molecule graphs shown in separate images, reading a rendered graph image altered by recoloring, patch perturbation, composition or attention focusing. Answer accuracy on the 100-item testmini split for this task, in percent; higher is better.
Metric: Answer accuracy (%; testmini). Source: arxiv.org. Saturation forecast: Around August 2027. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 30 | #82 |
| 2 | Qwen-VL-Max | 28 | #475 |
| 3 | Gemini 3 Flash | 27 | #103 |
| 4 | Grok 4 Fast | 25 | #266 |
| 5 | O1 | 24 | #190 |
| 6 | GPT-5.2 | 23 | #115 |
| 7 | GLM-4.6V | 18 | #330 |
| 8 | Kimi K2.5 (Thinking) | 10 | #152 (Kimi K2.5) |
| 9 | Qwen 3.5 35B A3B | 4 | #268 |
| 10 | Qwen 3.5 4B | 2 | #514 |
| 11 | InternVL3-8B | 1 | #624 |
| 12 | InternVL3-38B | 0 | #413 |
| 13 | QVQ-72B-Preview | 0 | #458 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=graphverse-graph-edit-distance-paired-images-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-05.