# google-research/magvit

Official JAX implementation of MAGVIT: Masked Generative Video Transformer

Repository: https://github.com/google-research/magvit
Canonical: https://ross.abutalabs.com/products/magvit
Homepage: https://magvit.cs.cmu.edu
Language: Python
License: Apache-2.0
License Family: permissive
Topics: generative-model, transformers, video-generation
Archived: true
Last push: 2024-01-17T18:06:44+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 98
- inputs: {"age_days": 1372, "days_push": 959, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1000, forks 47 (observed 2026-09-03T02:15:12.390138+00:00)

## What it is
Official JAX implementation of MAGVIT, a masked generative video transformer from a CVPR 2023 paper by Google Research and CMU. It provides code and pretrained models for a single model that handles many video synthesis tasks including prediction, interpolation, inpainting, outpainting, and text-to-video generation.

## Use cases
- generate videos from text prompts
- predict future video frames
- interpolate between video frames
- inpaint or outpaint videos
- convert images to animation
- research video generation transformers in JAX

## When to choose
- you need a research-grade JAX codebase for masked generative video modeling
- you want pretrained models for multi-task video synthesis like prediction, interpolation, and inpainting
- you are reproducing CVPR 2023 MAGVIT benchmark results

## When to avoid
- you need a production-ready or actively maintained video generation tool
- you prefer PyTorch rather than JAX
- you need simple APIs without deep learning research expertise

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, video-processing, image-processing, transformers
- domain: deep-learning, computer-vision, artificial-intelligence
- platform: python
- tags: jax, video-generation, video-tokenizer, masked-generative-model, research-code, cvpr-2023, text-to-video, frame-prediction, inpainting, outpainting, video, gpu, linux

## Member repositories
- google-research/magvit (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-09-03T02:15:12.390138+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 2026-08-30T07:16:06.433931+00:00, confidence not recorded.
  - readme: https://github.com/google-research/magvit (fetched 2026-09-03T02:15:12.390138+00:00, sha 03054129f293)
  - homepage: https://magvit.cs.cmu.edu (fetched 2026-08-29T13:14:10.747620+00:00, sha 294c7d052d76)
- Data as of 2026-08-30T08:39:29.467469+00:00.