NanoFLUX: Distillation-Driven Compression of Large Text-to-Image Generation Models for Mobile Devices

While large-scale text-to-image diffusion models continue to improve in visual quality, their increasing scale has widened the gap between state-of-the-art models and on-device solutions. To address this gap, we introduce NanoFLUX, a **2.4B** text-to-image flow-matching model distilled from **17B** FLUX.1-Schnell using a progressive compression pipeline designed to preserve generation quality. Our contributions include: (1) A model compression strategy driven by pruning redundant components in the diffusion transformer, reducing its size from 12B to 2B; (2) A ResNet-based token downsampling mechanism that reduces latency by allowing intermediate blocks to operate on lower-resolution tokens while preserving high-resolution processing elsewhere; (3) A novel text encoder distillation approach that leverages visual signals from early layers of the denoiser during sampling. Empirically, NanoFLUX generates $512 \times 512$ images in approximately 2.5 seconds on mobile devices, demonstrating the feasibility of high-quality on-device text-to-image generation.

State-of-the-art AI models capable of generating images from text have advanced considerably in recent years. However, their substantial size means they typically require powerful servers to operate, raising concerns around cost, accessibility and privacy. This study introduces NanoFLUX, a compressed AI image-generation model derived from the state-of-the-art model FLUX.1-Schnell. Through distillation, a process used to train a compact student model to replicate the performance of its much larger teacher, we reduced the system to roughly one-seventh of its original size while preserving image quality. Consequently, NanoFLUX is capable of generating high-quality images in approximately 2.5 seconds on mobile devices, without requiring an internet connection. We believe this work will serve as a methodological reference for distillation researchers and engineers in the future.