Figure AI introduces Helix VLA model for generalist humanoid control

Key points

• Figure introduced Helix as a Vision-Language-Action model that unifies perception, language understanding, and learned control for humanoid manipulation.
• The company says Helix outputs continuous control across the upper body, enables multi-robot collaboration, and runs on low-power embedded GPUs for commercial deployment.
• It matters because this is a first-party AI-system release that anchors Figure's research timeline to a concrete model rather than only hardware launches.