# Physical-Intelligence/openpi

Repository: https://github.com/Physical-Intelligence/openpi
Canonical: https://ross.abutalabs.com/products/openpi
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-08-24T16:27:13+00:00

## Health v2 (maintenance only)
Score: 66/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 35, longevity 48
- inputs: {"age_days": 681, "days_push": 9, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 13494, forks 2377 (observed 2026-08-28T04:11:03.408403+00:00)

## What it is
Open-source repository from Physical Intelligence containing vision-language-action (VLA) models for robotics, including π₀, π₀-FAST, and π₀.₅, with pre-trained checkpoints and fine-tuning examples. It provides packages for running inference and fine-tuning these models on custom robot datasets.

## Use cases
- run robot policy inference with pi0 on a custom robot
- fine-tune a vision-language-action model on my own robot dataset
- train a VLA on the DROID dataset
- use LoRA fine-tuning for pi0 on a single GPU
- evaluate open-source robot foundation model checkpoints
- adapt pi0 to ALOHA or other robot platforms

## When to choose
- you need state-of-the-art open VLA models for robot manipulation
- you have an NVIDIA GPU and want to fine-tune robot policies on your own data
- you want to experiment with flow-based or autoregressive action models on real robots

## When to avoid
- you need multi-node distributed training
- you work on non-Linux operating systems
- you lack a GPU with at least 8GB VRAM for inference

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-inference, llm-training
- domain: robotics, machine-learning, deep-learning, artificial-intelligence
- platform: python
- tags: vision-language-action, vla, robot-learning, foundation-models, fine-tuning, lora, flow-matching, pi0, pytorch, jax, linux, gpu

## Member repositories
- Physical-Intelligence/openpi (main) score 66

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:03.408403+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:13:15.794374+00:00, confidence not recorded.
  - readme: https://github.com/Physical-Intelligence/openpi (fetched 2026-08-28T04:11:03.408403+00:00, sha cd434f1b7322)
- Data as of 2026-08-30T08:39:29.467469+00:00.