List your company
ResearchFormerly WizWand

SVG Reconstruction on SVG

0.95Dino ScoreClaude 4.5 Sonnet
0.78360.82680.870.9132May 5, 2026
Updated 6mo ago

Evaluation results

MethodLinks
Claude 4.5 Sonnet2026.050.950.810.88
Claude 3.7 Sonnet2026.050.940.770.85
Gemini 2.5 Pro2026.050.930.780.86
GLANCE-FullBackbone=Qwen2.5-VL-3B, Reward Strategy=Dense Extrinsic, Reasoning Strategy=World Model Reasoning for Visual States2026.050.920.70.81
GLANCE-BaseBackbone=Qwen2.5-VL-3B, Reward Strategy=Sparse Extrinsic, Reasoning Strategy=World Model Reasoning Strategy2026.050.920.690.8
GRPO w/ MaskBackbone=Qwen2.5-VL-3B, Reward Strategy=RL Baseline, Reasoning Strategy=World Model Reasoning Strategy2026.050.920.660.79
GPT-4o2026.050.910.690.8
VAGEN-FullBackbone=Qwen2.5-VL-3B, Reward Strategy=Dense Extrinsic, Reasoning Strategy=World Model Reasoning for Visual States2026.050.90.660.78
VAGEN-BaseBackbone=Qwen2.5-VL-3B, Reward Strategy=Sparse Extrinsic, Reasoning Strategy=World Model Reasoning Strategy2026.050.90.650.78
GLANCE w/ Turn-PPOBackbone=Qwen2.5-VL-3B, Reward Strategy=Turn-level PPO, Reasoning Strategy=World Model Reasoning Strategy2026.050.90.660.78
o4-mini2026.050.90.660.78
Turn-PPO w/ MaskBackbone=Qwen2.5-VL-3B, Reward Strategy=Turn-level PPO, Reasoning Strategy=World Model Reasoning Strategy2026.050.890.640.77
Qwen2.5-VL-72BBackbone=Qwen2.5-VL-72B2026.050.840.620.73
Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.050.840.270.56
Vanilla-PPOBackbone=Qwen2.5-VL-3B, Reward Strategy=RL Baseline, Reasoning Strategy=World Model Reasoning Strategy2026.050.830.440.64
Qwen2.5-VL-3BBackbone=Qwen2.5-VL-3B2026.050.790.30.54
VLM-R1-3BBackbone=VLM-R1-3B2026.050.790.270.54