Robbyant Releases LingBot-VLA 2.0: An Open-Source 6B Vision-Language-Action (VLA) Model for Cross-Embodiment Robot Manipulation

Ant Group's Robbyant division has unveiled LingBot-VLA 2.0, a significant advancement in open-source artificial intelligence for robotic systems. This 6-billion-parameter vision-language-action (VLA) model represents a substantial step forward in developing AI systems capable of controlling robots across multiple hardware configurations. The release, distributed under the permissive Apache 2.0 license, aims to democratize access to sophisticated robotics AI technology and accelerate innovation in cross-embodiment robot manipulation.

LingBot-VLA 2.0 was pretrained on approximately 60,000 hours of comprehensive data, establishing a robust foundation for the model's capabilities. This training dataset comprises 50,000 hours of robot trajectories collected from 20 different robot configurations, alongside 10,000 hours of additional training data. The diverse hardware exposure enables the model to generalize across various robotic platforms, addressing a critical challenge in robotics development where most systems are optimized for specific hardware designs.

The open-source release positions LingBot-VLA 2.0 as a foundational resource for researchers and developers working on robotic applications. By providing a pretrained model with extensive cross-embodiment training, Robbyant removes significant barriers to entry for organizations developing robot manipulation systems.

• Enhanced accessibility to advanced robotics AI for smaller companies and research institutions
• Accelerated development timelines for new robotic applications through transfer learning capabilities
• Standardization opportunities in robot control across heterogeneous hardware environments
• Potential cost reduction in developing specialized robotics solutions
• Competitive advantages for organizations integrating language-based instruction systems into robotic workflows

The release of LingBot-VLA 2.0 signals the maturation of open-source approaches to robotics intelligence. Rather than relying on proprietary systems, the broader industry gains access to proven architectures trained on extensive real-world data. This democratization of robotics AI technology could reshape how organizations approach robot deployment, particularly in manufacturing, logistics, and research sectors. The cross-embodiment design ensures that improvements benefit the entire ecosystem rather than isolated applications, establishing a foundation for collaborative advancement in autonomous manipulation technology.