Figure AI

Figure AI is a humanoid robotics startup backed by OpenAI, founded in 2022 . Its first product, Figure 01, stands roughly 170 cm tall, weighs 60 kg, can carry 20 kg loads, and runs for five hours on a charge . The company has moved fast: by January 2024, Figure 01 learned to make coffee using end-to-end neural networks with just ten hours of training, and by February it was already moving boxes on conveyor belts, albeit at roughly one-sixth human speed . The March 2024 demo that drew wide attention showed Figure 01 conversing with humans and executing commands in real time after OpenAI's multimodal model was layered on top—an integration the two teams completed in thirteen days . Figure's architecture, as described by Yunqi Capital, stacks OpenAI's model for high-level vision and language reasoning atop its own neural networks for motion control and low-level execution . By August 2024, the company had already unveiled Figure 02, touting tripled on-board AI inference and real-time voice dialogue . Commercial traction began with a BMW manufacturing deployment in Spartanburg, South Carolina , and the company's stated ambition is to train a world model capable of selling humanoid robots at the billion-unit scale . In early 2025, Figure AI rolled out Helix, a vision-language-action (VLA) model aimed at generalizing across scene tasks .