GOVLA (Global & Omni-body Vision-Language-Action)

AI² Robotics' large VLA model that outputs both full-body actions and a movement trajectory at once, across arbitrary embodiments and spaces.

GOVLA was introduced by AI² Robotics in April 2025 alongside its AlphaBot 2 robot, at the same time the company upgraded its earlier embodied-model brand, AI2R Brain, to Alpha Brain, whose core is GOVLA (Global & Omni-body Vision-Language-Action model). It has three parts: a spatial-interaction foundation model, a slow system, and a fast system. The slow system (System 2) handles complex reasoning, task decomposition, and language interaction, while the fast system (System 1) outputs whole-body control actions and a movement trajectory. The company's framing is that ordinary VLA models only output robot-arm actions and can only operate within tabletop range, whereas GOVLA targets tasks ranging from the tabletop to open environments and from single-arm to whole-body coordination, and incorporates technology from DeepSeek in its construction to strengthen long-horizon reasoning.

ExampleAs the company's own example puts it: an ordinary VLA robot needs a person to first put the ingredients on the table before it can work, whereas a robot running GOVLA can go get ingredients from the refrigerator itself, make breakfast, and bring it to the table.

Also called

Alpha Brain, AlphaBrain, AI2R Brain

Related

AI² Robotics · AI² Robotics AlphaBot · Dual-System Architecture (System 1 / System 2) · Vision-Language-Action Model · RoboMamba · Mobile Manipulation

Sources

智平方发布全新一代智能机器人AlphaBot 2，开启AGI终端新时代！（智平方官网） (Chinese)

As of

2025-04