VisualThink-VLA: Visual Intermediate Reasoning for Effective and Low-Latency Vision-Language-Action Policies

Benchmark Model Rank Results
robot-manipulation-on-libero-longVisualThink-VLA#39Success Rate: 95.87
robot-manipulation-on-libero-goalVisualThink-VLA#56Success Rate: 97.05
robot-manipulation-on-libero-spatialVisualThink-VLA#79Success Rate: 96.69
robot-manipulation-on-libero-objectVisualThink-VLA#81Success Rate: 97.74