one year on
Figure 01 humanoid robot demonstrates natural conversation and tasks with OpenAI-powered reasoning
In a video released today, the Figure 01 robot speaks, explains its actions, and performs household tasks in real-time, marking a major step forward for embodied AI.
Figure, the humanoid robotics startup that emerged from stealth last year, today released a video demonstrating its Figure 01 robot performing household tasks while engaging in natural conversation with a human operator. The robot identifies objects, explains its reasoning, and manipulates items with dexterity, all in real-time.
The demo shows the robot picking up an apple when asked for food, explaining that it is the only edible item available, and then placing dishes into a drying rack. Figure and OpenAI partnered less than two weeks ago to integrate OpenAI’s vision-language and speech models with Figure’s neural networks. Figure co-founder Brett Adcock confirmed on X that the robot operates via end-to-end neural networks with no teleoperation, filmed at 1.0x speed.
The video has drawn comparisons to Tesla’s Optimus robot, which critics note has been shown in teleoperated demos. The demo follows Figure’s announcement of a successful Series B funding round and a collaboration agreement with OpenAI.
The record
Emphasized that the video shows end-to-end neural networks with no teleoperation, filmed at 1.0x speed and shot continuously.
Stated that all behaviors are learned, not teleoperated, and run at normal speed. Explained that the system uses images from robot cameras and transcribed speech fed into a large multimodal model trained by OpenAI.
One year later — open only if you can handle spoilers
Figure 01's demo became a viral sensation, but the company faced challenges scaling production and deploying in factories. By mid-2026, Figure had partnered with several manufacturers but had not achieved the widespread commercial deployment predicted. The demo nonetheless raised the bar for humanoid robotics and spurred investment in the sector.