HunyuanImage 3.0 Technical Report

2025-09-28

AI Summary

This work addresses the challenge of unifying multimodal understanding and generation within a single autoregressive framework. To this end, we propose a native chain-of-thought architecture and a progressive training strategy, enabling—for the first time in open-source image generation—the successful training and deployment of an over-80-billion-parameter Mixture-of-Experts (MoE) model, with only 13 billion parameters activated per inference step. Leveraging large-scale high-quality multimodal data curation, distributed training system optimizations, and aggressive post-training techniques, we significantly improve text–image alignment and generative visual fidelity. Comprehensive human and automated evaluations demonstrate performance on par with state-of-the-art proprietary models. All code, model weights, and detailed training configurations are publicly released, establishing a foundational infrastructure and introducing a novel paradigm for multimodal foundation model research.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Large Vision Models
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

We present HunyuanImage 3.0, a native multimodal model that unifies multimodal understanding and generation within an autoregressive framework, with its image generation module publicly available. The achievement of HunyuanImage 3.0 relies on several key components, including meticulous data curation, advanced architecture design, a native Chain-of-Thoughts schema, progressive model pre-training, aggressive model post-training, and an efficient infrastructure that enables large-scale training and inference. With these advancements, we successfully trained a Mixture-of-Experts (MoE) model comprising over 80 billion parameters in total, with 13 billion parameters activated per token during inference, making it the largest and most powerful open-source image generative model to date. We conducted extensive experiments and the results of automatic and human evaluation of text-image alignment and visual quality demonstrate that HunyuanImage 3.0 rivals previous state-of-the-art models. By releasing the code and weights of HunyuanImage 3.0, we aim to enable the community to explore new ideas with a state-of-the-art foundation model, fostering a dynamic and vibrant multimodal ecosystem. All open source assets are publicly available at https://github.com/Tencent-Hunyuan/HunyuanImage-3.0

Problem

• Unifies multimodal understanding and generation in autoregressive framework
• Develops efficient infrastructure for large-scale training and inference
• Creates open-source image generation model rivaling state-of-the-art performance

Innovation

• Unifies multimodal understanding and generation autoregressively
• Employs Mixture-of-Experts with 80B total parameters
• Uses progressive pre-training and aggressive post-training