The brief

Analyst view

• Dyna-2.1: 200 units/hour at ±2mm accuracy; 22-min hotel room turnover; 100 dishes/hour service.
• Raised $23.5M seed (March 2025) and $120M Series A (September 2025) from CRV, First Round, RoboStrategy, NVentures, Amazon, Samsung and LG.
• Launched commercially September 2026 with deployments in hotels, restaurants and laundromats.

Technical approach

As reported

Modular base

Wheeled mobile base with steerable wheels

Unlike legged humanoids, Dyna-2.1 uses four steerable wheels, eliminating balance requirements and enabling stable operation on varied surfaces.

Dexterous manipulation

Dual arms with flexible end-effectors

Two arms can be fitted with parallel-jaw grippers for simple pick-and-place or dexterous hands for complex manipulation, depending on the task requirements.

Vision-language integration

Multi-modal understanding of tasks

Vision-language models with policy and world-action modeling understand and execute complex task sequences.

Embodied learning

Reinforcement learning in simulation and real world

Models trained in simulation with multi-sensory feedback to handle varied conditions and novel objects.

Show all 4