Alpha-VLLM/Lumina-DiMOO

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1015 stars · 63 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Lumina-DiMOO is an open-source omni diffusion large language model that uses fully discrete diffusion to handle multimodal inputs and outputs. It supports text-to-image generation, image-to-image tasks like editing and inpainting, and image understanding, with training and inference code and checkpoints released.

Use cases

• generate images from text prompts with a diffusion language model
• edit images or do subject-driven generation and inpainting
• understand and answer questions about images with a multimodal model
• research discrete diffusion models for multimodal learning
• speed up multimodal sampling with caching and test-time scaling
• fine-tune a unified multimodal model with GRPO-style training

When to choose

• you need a unified model for both multimodal understanding and generation
• you want faster sampling than autoregressive or hybrid AR-diffusion models
• you are researching discrete diffusion language models
• you want an Apache-2.0 licensed open multimodal foundation model

When to avoid

• you need a lightweight model for CPU-only or edge deployment
• you only need a production chatbot with mature ecosystem tooling
• you require long-form text-only generation with standard AR LLMs

Facets

library · maturity active

machine-learning deep-learning image-processing llm-inference llm-training artificial-intelligence large-language-models image-processing computer-vision deep-learning python discrete-diffusion multimodal text-to-image image-editing diffusion-language-model image-understanding gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Alpha-VLLM/Lumina-DiMOO")