# FoundationVision/LlamaGen

Autoregressive Model Beats Diffusion: 🦙 Llama for Scalable Image Generation

Repository: https://github.com/FoundationVision/LlamaGen
Canonical: https://ross.abutalabs.com/products/llamagen
Homepage: https://arxiv.org/abs/2406.06525
Language: Python
License: MIT
License Family: permissive
Topics: auto-regressive-model, diffusion, diffusion-models, image-generation, llama, llm, text2image
Last push: 2024-08-15T20:10:35+00:00

## Health v2 (maintenance only)
Score: 24/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 58
- inputs: {"age_days": 821, "days_push": 748, "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 1966, forks 95 (observed 2026-08-28T04:05:59.907370+00:00)

## What it is
LlamaGen is a family of autoregressive image generation models that apply next-token prediction, as in Llama-style LLMs, to visual generation. The repo provides PyTorch training/sampling code, image tokenizers, and pre-trained class- and text-conditional models from 100M to 3B parameters, with vLLM serving support.

## Use cases
- generate images from text prompts
- class-conditional image generation on ImageNet
- train an autoregressive text-to-image model
- speed up image generation inference with vLLM
- research next-token prediction for visual generation
- compare autoregressive models against diffusion models

## When to choose
- you want to experiment with autoregressive image generation instead of diffusion
- you need pre-trained image generation models with open weights
- you are researching scalable visual token prediction

## When to avoid
- you need a production-ready text-to-image service with broad ecosystem support
- you lack GPU resources for large model inference
- you want actively developed tooling with frequent updates

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, image-processing, llm-inference
- domain: deep-learning, computer-vision, image-processing, artificial-intelligence
- platform: python
- tags: autoregressive-model, text-to-image, vq-vae, image-tokenizer, pytorch, diffusion-alternative, vllm, pretrained-models, gpu

## Member repositories
- FoundationVision/LlamaGen (main) score 24

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:59.907370+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 2026-08-30T03:05:26.144789+00:00, confidence not recorded.
  - readme: https://github.com/FoundationVision/LlamaGen (fetched 2026-08-28T04:05:59.907370+00:00, sha b1c7078f1281)
  - homepage: https://arxiv.org/abs/2406.06525 (fetched 2026-08-29T10:44:59.593873+00:00, sha 8fcc650a1b0d)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T10:44:59.603208+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T10:44:59.606752+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T10:44:59.608526+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T10:44:59.605061+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.