Skip to content

Figure AI Helix

A vision language action model for humanoid robots

Visit official site

What Figure AI Helix is

The model is described as a vision language action model for general humanoid robot control, joining visual perception, language understanding and action control. It aims at immediate behavior generation for home environments.

What you can do with it

  • Control the whole upper body including fingers
  • Run two robots on one shared task
  • Grasp unfamiliar items from spoken style commands

Getting started

  • Prepare a robot with a low power embedded GPU
  • Deploy the model and load pretrained weights
  • Issue a natural language instruction
  • Observe the generated behavior

Pros & cons

✓ What we like

  • Natural language commands handle unfamiliar small objects
  • It runs on low power embedded GPUs

! What to watch out for

  • A robot with an embedded GPU is needed
  • Pretrained network weights must be loaded first

FAQ

Does it need task specific fine tuning?

The material describes a single neural network that needs no task specific fine tuning.

Can two robots work together?

Yes, two robots using the same model weights on a shared task are listed.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review