VCG-Bench - Diagram Editing - XDRFR: leaderboard
Metric: Requirements Following Ratio (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro | 94.05 |
| 2 | GPT-5.2 | 93.34 |
| 3 | Qwen 3 VL 235B A22B | 93.12 |
| 4 | Gemini 3 Flash | 92.95 |
| 5 | Claude Sonnet 4.5 | 92.87 |
| 6 | Kimi K2 | 92.76 |
| 7 | Claude Opus 4.5 | 92.71 |
| 8 | Qwen3 Coder | 92.1 |
| 9 | GLM-4.6 | 92.02 |
| 10 | MiniMax-M2 | 91.51 |
| 11 | DeepSeek V3.2 | 89.28 |
Interactive version: theaggregate.ai/benchmark?slug=vcg-bench-diagram-editing-xdrfr · How It Works · Data refreshed daily, snapshot 2026-09-25.