List your company
ResearchFormerly WizWand

Robotic Manipulation on PrimitiveSkill

100Place Success RateQwen2.5-VL-72B
-4235077May 5, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Qwen2.5-VL-72BBackbone=Qwen2.5-VL-72B2026.0510050010044
VAGEN-FullBackbone=Qwen2.5-VL-3B, Reward Strategy=Dense Extrinsic, Reasoning Strategy=World Model Reasoning for Visual States2026.051008810010097
GLANCE-FullBackbone=Qwen2.5-VL-3B, Reward Strategy=Dense Extrinsic, Reasoning Strategy=World Model Reasoning for Visual States2026.051008810010097
VAGEN-BaseBackbone=Qwen2.5-VL-3B, Reward Strategy=Sparse Extrinsic, Reasoning Strategy=World Model Reasoning Strategy2026.0510088888891
GLANCE-BaseBackbone=Qwen2.5-VL-3B, Reward Strategy=Sparse Extrinsic, Reasoning Strategy=World Model Reasoning Strategy2026.05100881008894
GLANCE w/ Turn-PPOBackbone=Qwen2.5-VL-3B, Reward Strategy=Turn-level PPO, Reasoning Strategy=World Model Reasoning Strategy2026.0510063010066
o4-mini2026.051005007556
Gemini 2.5 Pro2026.05636307550
Claude 4.5 Sonnet2026.056350010053
Claude 3.7 Sonnet2026.056313010044
GPT-4o2026.05506308850
Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.050007519
Qwen2.5-VL-3BBackbone=Qwen2.5-VL-3B2026.0500000
VLM-R1-3BBackbone=VLM-R1-3B2026.0500000
Turn-PPO w/ MaskBackbone=Qwen2.5-VL-3B, Reward Strategy=Turn-level PPO, Reasoning Strategy=World Model Reasoning Strategy2026.0500010025
Vanilla-PPOBackbone=Qwen2.5-VL-3B, Reward Strategy=RL Baseline, Reasoning Strategy=World Model Reasoning Strategy2026.0500000
GRPO w/ MaskBackbone=Qwen2.5-VL-3B, Reward Strategy=RL Baseline, Reasoning Strategy=World Model Reasoning Strategy2026.0500010025