What is Figure Helix 2.5?

Robots that work in homes face a hard problem: teaching a robot even one new behavior takes a great deal of human effort, and Figure says robots usually learn the places they work in one place at a time. Helix 2.5 is a neural network (a learning program) from Figure, built to control a humanoid robot's whole body. The maker says it was pretrained (first trained broadly, before being adapted) on Index, its dataset of human behavior, and then adapted to three household behaviors: tidying living rooms, folding towels and making beds. Its stated aim is to let a humanoid enter a home it has never seen and start working without being retrained there. Figure listed it on September 17, 2026.

Written from primary sources and reference pages, and checked against them by an independent AI fact-checker. Drafted with AI assistance.