VectorGym: leaderboard
Metric: Overall VectorGym score (0-100): unweighted mean of the Sketch2SVG, SVG Editing, Text2SVG and SVG Captioning task scores on VectorGym's human-annotated test split (real-world SVGs from SVG-Stack); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 73.17 | #77 |
| 2 | GPT-5.1 | 71.36 | #131 |
| 3 | Claude Sonnet 4.5 | 70.31 | #138 |
| 4 | GPT-4o | 64.93 | #333 |
| 5 | Qwen 3 VL 235B A22B Instruct | 62.32 | #264 |
| 6 | Gemini 2.5 Flash | 61.42 | #237 |
| 7 | Qwen 3 VL 8B Instruct | 58.74 | #401 |
| 8 | GLM-4.5V | 57.02 | #339 |
| 9 | Qwen 2.5 VL 32B Instruct | 49.16 | #443 |
| 10 | Qwen 2.5 VL 72B Instruct | 44.27 | #364 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=vectorgym · How It Works · Data refreshed daily, snapshot 2026-10-11.