Generalist Releases GEN-1 Robotic Intelligence Foundation Model

Generalist released GEN-1, a new embodied-robotics foundation model that claims mastery-level performance on simple physical tasks. GEN-1 emphasizes three engineering goals — reliability, speed and improvisation — and the company reports task success rates exceeding 99%, motion execution up to ~3× faster than prior art, and improved recovery from interruptions. Example benchmarks cited include assembling a box in 12.1 seconds versus 34 seconds for prior models. Generalist positions GEN-1 as a step toward scaling ‘mastery’ across diverse physical work by training on raw movement data and optimizing inference kernels and attention mechanisms for real-time control.