PhyAI - Latency-first serving for Physical AI

Open-source serving engine

Latency first / Physical AI

Latency-first serving for Physical AI.

Run policy and action models where response time is part of correctness.

pi0.5 / single GPU

Python

from phyai import Engine, EngineArgs
from phyai.models.pi05.main_pi05 import PI05Args

engine = Engine(EngineArgs(
    plugin="pi05",
    plugin_args=PI05Args(
        checkpoint_dir="lerobot/pi05_base",
    ),
))
actions = engine.step(request)
engine.close()

Nightly build

uv pip install phyai phyai-ext \
  --extra-index-url https://membodied.github.io/phyai/simple/ \
  --prerelease=allow

Introducing PhyAI: Real-Time Physical AI at the Edge, Scalable Rollouts in the Cloud

Read the post