Meet Qwen-RobotSuite: Three Embodied AI Models for VLA Manipulation, Video World Modeling, and Navigation

Short summary

Qwen releases RobotSuite, a suite of three new embodied AI models for autonomous robotic control and manipulation. RobotManip is a Vision-Language-Action model built on Qwen3.5-4B for manipulation, while RobotWorld combines language-conditioning with a 60-layer MMDiT for video world modeling. RobotNav handles autonomous navigation across 2B to 8B model sizes with comprehensive architectural and benchmark documentation.

• •Three new embodied AI models: RobotManip (manipulation), RobotWorld (video modeling), RobotNav (navigation)
• •Built on Qwen3.5-4B and Qwen3-VL with model sizes ranging from 2B to 8B parameters
• •Covers full technical pipeline: architecture, training data, and benchmark results for each model

Generated with AI, which can make mistakes.