Swift-Image: A Compact 6B Unified Model for Generation and Editing at the Performance Frontier

Forum topic · 小凯 · 2026-08-22

Summary

Swift-Image (arXiv:2608.20334) is a compact unified model covering text-to-image generation, single-image editing, and multi-image editing. Built on an efficient 6B single-stream Diffusion Transformer (DiT), the project explores how far relatively small visual generators can be pushed under constrained compute budgets through systematic training engineering. A progressive training pipeline evolves from broad semantic coverage to higher resolution, stronger visual quality, and unified generation-editing supervision. Post-training applies parallel-expert reinforcement learning followed by multi-teacher on-policy distillation to reduce interference among heterogeneous objectives. A Prompt Enhancer translates user requests into generator-aligned visual specifications, decoupling high-level reasoning from pixel-level rendering. Structural pruning and few-step distillation yield 3B and accelerated variants. With only 6B parameters and 243K GPU training hours, Swift-Image reports leading comprehensive performance among evaluated open-source models; the compressed 3B model shows near-lossless quality, and few-step distillation further improves editing performance with far fewer sampling steps.

Paper Overview

Field: Computer Vision (CV)
Authors: Taihang Hu, Zhao Wang, Zuan Gao, Tao Liu
Released: 2026-08-22
arXiv: 2608.20334

Abstract

Swift-Image is a compact unified model for text-to-image generation, single-image editing, and multi-image editing. The goal is to explore, through systematic training engineering, how far a relatively small visual generator can be pushed under a constrained compute budget.

Key components:

Efficient 6B single-stream DiT with a progressive training pipeline that evolves 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 to mitigate interference among heterogeneous objectives.
Prompt Enhancer: translates user requests into generator-aligned visual specifications, decoupling high-level reasoning from pixel-level rendering.
Compression: structural pruning and few-step distillation produce 3B and accelerated variants.

Results

With only 6B parameters and 243K GPU training hours, Swift-Image achieves leading comprehensive performance among evaluated open-source models.
The compressed 3B model is nearly lossless.
Few-step distillation further improves comprehensive editing performance with substantially reduced sampling steps.

---

*Auto-collected on 2026-08-22.*