VLAnew

Pi-0

Physical Intelligence · 3.3B · Bimanual · 50Hz

Pi-0 ships strong zero-shot bimanual priors. Fine-tune on your teleop dataset and serve on either our sim cluster or your robots over WebRTC. Pricing covers the full sim+real loop.

Train this model →Read the recipe

Quickstart

Launch a Pi-0 fine-tune

~ fourbit

$ fourbit train \
  --base pi-zero \
  --episodes s3://acme/teleop/ \
  --recipe fullFt

More VLAs

OpenVLA 7B
Open-source 7B VLA. Solid baseline for single-arm manipulation tasks.
7B