Generalist AI

Generalist AI is a San Mateo, California company founded in 2024 by Pete Florence, its CEO, Andy Zeng, chief scientist, and Andrew Barry, CTO. It does not build robots. Its business is training what it calls embodied foundation models on very large sets of real manipulation data and licensing them to run on other people's arms, hands and humanoids.

The first public model, GEN-0, came out in November 2025 with a claim that robot learning follows scaling laws like language models. GEN-1 followed in April 2026, trained from scratch on what the company says is half a million hours of real-world data, with a claimed 99 percent success rate on tasks where earlier models reached 64 percent and about an hour of robot data needed per new task. GEN-1.5, announced on 19 August 2026, is pitched as a one-shot learner from as little as 12 seconds of demonstration. These numbers are the company's own.

On 4 June 2026 it announced a 400 million US dollar round led by Radical Ventures, with 8VC, Union Square Ventures, Hanabi Capital and Norwest joining and Nvidia's NVentures, Boldstart, Spark Capital, Bezos Expeditions and NFDG returning, taking total funding past 500 million US dollars. It has named no paying customers or deployments, and no headcount or valuation has been published.