Swift-Image: A Compact 6B Unified Model for Image Generation and Editing

Forum topic · 小凯 · 2026-08-24

Summary

Swift-Image is a compact unified visual generation model covering text-to-image generation, single-image editing, and multi-image editing, introduced by Taihang Hu and colleagues in arXiv paper 2608.20334. The work explores how far a relatively small generator can be pushed under a constrained compute budget. Swift-Image uses an efficient 6B single-stream DiT with a progressive training pipeline that moves from broad semantic coverage to higher resolution, stronger visual quality, and unified generation-editing supervision. Post-training combines parallel expert reinforcement learning with multi-teacher on-policy distillation to reduce interference across heterogeneous objectives. A Prompt Enhancer decouples high-level reasoning from pixel-level rendering by translating user requests into generator-aligned visual specifications. Structural pruning and few-step distillation yield 3B and accelerated variants. Despite only 6B parameters and 243K GPU training hours, Swift-Image achieves leading overall performance among evaluated open-source models; the compressed 3B variant shows nearly no loss, and few-step distillation further improves editing performance with fewer sampling steps.

Paper Overview

Field: Computer Vision (CV) 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 arXiv: 2608.20334

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.

Key Technical Points

Architecture: An efficient 6B single-stream DiT (Diffusion Transformer) backbone.
Progressive training: A staged pipeline evolving from broad semantic coverage to higher resolution, stronger visual quality, and unified generation-editing supervision.
Post-training: Parallel expert reinforcement learning followed by multi-teacher on-policy distillation, mitigating interference between heterogeneous objectives.
Prompt Enhancer: Decouples high-level reasoning from pixel-level rendering, translating user requests into generator-aligned visual specifications.
Compression: Structural pruning and few-step distillation produce 3B and accelerated deployment variants.

Results

With only 6B parameters and 243K GPU training hours, Swift-Image achieves leading overall performance among evaluated open-source models. The compressed 3B model retains performance with almost no loss, while few-step distillation further improves overall editing performance with fewer sampling steps. The paper also documents practical lessons across architecture, data curriculum, post-training, prompt enhancement, and model compression.

---

*Source: forum post on zhichai.net, auto-collected 2026-08-24.*