Swift-Image: Exploring the Performance Frontier of Compact Unified Image Generation Models
2026-08-20 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors created Swift-Image, a compact model that can generate images from text and edit single or multiple images. They focused on making the model efficient and small by carefully designing the training process and model structure. To improve the model's abilities, they used special training methods and a component that helps convert user instructions into clear visual details. They also made smaller, faster versions of the model without much loss in quality. Overall, their work shows how to build a versatile image model using limited resources.
text-to-image generationimage editingmodel compressionreinforcement learningdistillationDiT (Diffusion Transformer)prompt enhancementtraining pipelinemodel pruningsemantic coverage
Authors
Taihang Hu, Zhao Wang, Zuan Gao, Tao Liu, Hao Yan, Zhengze Xu, Yuhang Yu, Yongchao Du, Xingjian Wang, Jun Zheng, Qinye Zhou, Zhengrui Chen, Chao Lin, Yefeng Shen, Zhengtao Wu, Ge Wu, Xiaoli Xu, Denghui Yang, Huayu Zhang, Mingzhou Zhang, Mengting Chen
Abstract
We present Swift-Image, a compact unified model for text-to-image generation, single-image editing, and multi-image editing. Our goal is to explore how far a relatively small visual generator can be pushed through systematic training engineering under a constrained computational budget. Swift-Image adopts an efficient 6B single-stream DiT and a progressive training pipeline that evolves from broad semantic coverage to higher resolution, stronger visual quality, and unified generation-editing supervision. For post-training, we employ parallel expert reinforcement learning followed by multi-teacher on-policy distillation to alleviate interference among heterogeneous objectives. We further decouple high-level reasoning from pixel-level rendering with a Prompt Enhancer that translates user requests into generator-aligned visual specifications. For efficient deployment, structural pruning and few-step distillation produce 3B and accelerated variants. Swift-Image achieves leading aggregate performance among evaluated open-source models with only 6B parameters and 243K GPU training hours; the compressed 3B model incurs nearly no loss, while few-step distillation further improves aggregate editing performance with substantially fewer sampling steps. Our study also summarizes practical lessons for architecture, data curriculum, post-training, prompt enhancement, and model compression.