# google-research/parti

Repository: https://github.com/google-research/parti
Canonical: https://ross.abutalabs.com/products/parti
License: Apache-2.0
License Family: permissive
Last push: 2022-06-28T23:02:25+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 1539, "days_push": 1527, "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 1589, forks 85 (observed 2026-08-28T04:05:08.141328+00:00)

## What it is
Parti is Google Research's Pathways Autoregressive Text-to-Image model, which generates high-fidelity photorealistic images from text by treating generation as sequence-to-sequence modeling over image tokens produced by ViT-VQGAN. The repository also releases PartiPrompts (P2), a benchmark of 1600+ English prompts for evaluating text-to-image models.

## Use cases
- generate photorealistic images from text prompts
- research autoregressive text-to-image generation
- benchmark text-to-image models with PartiPrompts
- encode and reconstruct images as discrete tokens with ViT-VQGAN
- study scaling behavior of large generative vision-language models

## When to choose
- you are researching autoregressive text-to-image generation or model scaling
- you need a holistic prompt benchmark for evaluating text-to-image models
- you want to build on ViT-VQGAN image tokenization

## When to avoid
- you need a production-ready image generation service or API
- you want a diffusion-based generator like Stable Diffusion or Imagen
- you lack large-scale GPU/TPU resources for 20B-parameter models

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing
- domain: artificial-intelligence, machine-learning, image-processing
- platform: python
- tags: text-to-image, autoregressive-model, image-generation, research, benchmark, vit-vqgan, google-research, gpu, linux

## Member repositories
- google-research/parti (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:08.141328+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 2026-08-30T03:54:50.606382+00:00, confidence not recorded.
  - readme: https://github.com/google-research/parti (fetched 2026-08-28T04:05:08.141328+00:00, sha bb78764d6b2a)
- Data as of 2026-08-30T08:39:29.467469+00:00.