Embodied foundation models for robots across tasks and form factors
HQ: San Mateo, California, United States; Bay Area and Boston operations | Founded: 2024 | Employees: Not disclosed in the reviewed sources | Stage: $400M June 2026 financing; approximately $200M extension reported August 2026 | Website: https://generalistai.com/
Generalist develops embodied foundation models with an initial focus on dexterity. Its official about page describes a team with experience at OpenAI, Boston Dynamics, and Google DeepMind and operations in the Bay Area and Boston. The Robot Report identifies San Mateo as its base and 2024 as its founding year.
GEN-0 was introduced in November 2025 and GEN-1 in April 2026. In June materials Generalist claimed GEN-1 achieved 99% reliability on evaluated tasks and up to 3x faster execution than prior state of the art. These company-reported results should not be extrapolated to all industrial environments. August reporting describes GEN-1.5 learning from short demonstrations and work with a handful of unnamed customers.
June funding was led by Radical Ventures, with new and existing investors supporting models, physical data collection, compute, and deployment work. August reporting describes additional capital led by 8VC. Complete cumulative financing, customer contracts, and revenues were not established in primary materials.
Generalist is a robot-intelligence company, not a manufacturer selling a single robot body. Its thesis is that large-scale physical experience can train models useful across tasks and form factors. The founders combine DeepMind robotics research and Boston Dynamics engineering, relevant experience for bridging model capability and physical deployment.
The company announced $400M in June 2026, taking confirmed cumulative financing above $500M. Axios subsequently reported about $200M more; TechCrunch reporting syndicated by Yahoo described a $3B valuation. The June round's label is inconsistent: Radical's portfolio page calls it Series A, while August TechCrunch coverage calls it Series B. This profile preserves that discrepancy rather than inventing a resolution.
Editorial assessment: hardware-spanning intelligence could serve many robot businesses, but reliable transfer to customer environments remains the challenge. Generalist's 99% reliability and speed figures are scoped company benchmark claims, not a safety or reliability guarantee across all tasks.
Category: Robotics & Automation Startups
Browse: Home | Blog | About | All Startups A-Z | Robotics & Automation Startups | View full profile on StartupWiki