Ant Group's LingBot-VLA 2.0 Goes Open Source, Supporting 17 Robot Manufacturers

On July 8, Ant Group's robotics company Ant Lingbo announced the open-source release of LingBot-VLA 2.0, its vision-language-action foundation model for robots. This follows the initial 1.0 release earlier this year.

LingBot-VLA integrates visual perception, language understanding, and motion control into a unified architecture, enabling robots to interpret natural language commands and execute actions directly. The model is designed as a general-purpose brain for diverse robotic hardware.

According to Ant Lingbo, version 2.0 delivers improvements in three key areas: stronger generalization across environments, enhanced motion control precision, and broader multi-configuration compatibility. The model now supports over 20 robot configurations from 17 manufacturers, covering humanoid robots, quadruped robots, and robotic arms.

This open-source move signals Ant Group's strategy to position its robotics AI as infrastructure for the entire industry, rather than keeping it proprietary. Developers can freely customize and fine-tune LingBot-VLA 2.0, lowering the barrier to entry for embodied AI applications.

What sets LingBot-VLA 2.0 apart is its cross-platform compatibility. In the embodied AI space, hardware differences across manufacturers are substantial. A VLA model that works across multiple hardware platforms can significantly reduce redundant development costs and push the industry toward scalable deployment.

Ant Group's robotics ambitions are increasingly moving from lab research toward industrial application. By open-sourcing LingBot-VLA, Ant is positioning itself as an AI infrastructure provider for robotics, similar to the open-source strategies seen in the large language model ecosystem.

The key question ahead is whether LingBot-VLA 2.0 can attract enough developer and manufacturer adoption to create a network effect, and how competitors in the embodied AI race will respond to this open-source push.