# LLaDA2.0-Uni — Unified Discrete Diffusion LLM for Multimodal Understanding and Generation

> One 16B discrete diffusion model that both understands and generates images — no separate encoder or decoder head.

- Org: Inclusion AI
- Date: 2026-04-22
- Category: paper, algorithm, model
- Source: https://arxiv.org/abs/2604.20796
- Canonical page: https://ai-tldr.dev/releases/inclusionai-llada2-0-uni/

## In short

#1 HuggingFace paper today (129 upvotes). A 16B MoE discrete diffusion model that handles multimodal understanding AND image generation in one architecture using masked token prediction — no separate specialized components. Apache 2.0.

## What it is

LLaDA2.0-Uni is a 16B Mixture-of-Experts discrete diffusion language model from Inclusion AI's AGI Research Center that unifies multimodal understanding (visual QA, captioning, document reading) and image generation (text-to-image, image editing) in a single architecture. Released April 22, 2026 with weights on HuggingFace and code on GitHub. Apache 2.0 license.

## How it works

Visual inputs are tokenized into discrete semantic tokens via SigLIP-VQ, converting continuous pixel information into a vocabulary compatible with the discrete diffusion backbone. The MoE backbone applies block-level masked diffusion to both text and vision tokens uniformly. A lightweight diffusion decoder (trained with few-step distillation) converts output discrete tokens back to pixels in ~8 steps. The SPRINT acceleration system reuses KV caches and adaptively unmasks tokens to reduce inference cost.

## Why it matters

Most multimodal systems chain separate specialized models — a VLM for understanding, a diffusion model for generation. LLaDA2.0-Uni collapses this into a single model trained with one objective, which simplifies pipelines and enables interleaved reasoning-then-generation tasks without model switching. The model matches specialized VLMs on understanding benchmarks while retaining strong image generation quality.

## Who it's for

ML researchers studying unified multimodal architectures; practitioners building image generation or editing pipelines.

## Try it

pip install transformers && model = AutoModelForCausalLM.from_pretrained('inclusionAI/LLaDA2.0-Uni', trust_remote_code=True)

## Links

- [Paper](https://arxiv.org/abs/2604.20796)
- [Model](https://huggingface.co/inclusionAI/LLaDA2.0-Uni)
- [Code](https://github.com/inclusionAI/LLaDA2.0-Uni)