This is Figure’s humanoid robot, which can now use OpenAI’s large vision language model (VLM) to provide reasoning and language understanding. The video shows how the bot can identify and interact with the objects on the counter in front of it when given a prompt, like “Can I have something to eat?”