"But is this brute-force scaling truly the only path forward?"
> > *Using general-purpose image inpainting as our strategic entry point, we challenge the "scale-at-all-costs" path dependency dictated by the Scaling Law narrative. Through the synergistic optimization of architectural design and knowledge distillation, Moebius achieves a remarkably compact footprint of just **0.22B parameters**. It liberates high-quality image inpainting from the heavy-compute narrative of 10B+ foundation models:* > *Across six comprehensive benchmarks spanning both natural and portrait scenes, Moebius performs **on par with**, and in certain scenarios **surpasses**, the inpainting quality of 10B+ industrial state-of-the-art (SOTA) generalist models like *FLUX.1-Fill-Dev*, while delivering a massive **>15ร inference acceleration**.* > > ๐ก **The core insight of Moebius can be summarized in a single equation:** > > $$\begin{aligned} > \text{Synergy} \times (\text{Architecture} + \text{Distillation}) = & \text{Shattering the "Impossible Triangle" of} \\ > & \text{Low Parameters, Fast Inference, and High Quality} > \end{aligned}$$ > > --- *written on June 16, 2026* --- ## ๐ Highlights - **๐ Extreme Parametric Efficiency (< 2%)**: Moebius operates with a mere **0.22B (226M) parameters**, which represents **less than 2%** of the size of the colossal industrial giant *FLUX.1-Fill-Dev (11.9B)*. It shatters the heavy-compute narrative, making high-quality inpainting accessible on consumer-grade and edge devices. - **โก 15ร Inference Speedup (26ms/step)**: Achieves a blistering inference latency of only **26.01 ms per step** on a single GPU. Combined with optimized sampling steps, Moebius delivers an overall **>15ร total runtime acceleration** compared to 10B-level models. - **๐ 10B-Level Inpainting Quality (on-par-with/surpass FLUX.1-Fill-Dev across 6 benchmarks)**: Size contraction does not mean representation degradation. Through the synergistic optimization of architecture and distillation, Moebius performs on par with, and in certain scenarios (such as complex textures and facial plausibility), surpasses 10B-level state-of-the-art (SOTA) generalist models (*FLUX.1-Fill-Dev, SD3.5 Large-Inpainting*) across 6 comprehensive benchmarks spanning **both natural** scenes (*Places2*) and **portrait** scenes (*CelebA-HQ*, *FFHQ*). - **๐ก Synergistic Core Innovations**: - **Architecture Design (LฮปMI Block)**: Reformulates both self- and cross-attention by condensing spatial context and global semantic priors into fixed-size linear matrices, bypassing quadratic computational overhead. - **Adaptive Multi-Granularity Distillation Strategy**: Transfers the representational capacity from our *[PixelHacker](https://github.com/hustvl/PixelHacker)* (teacher) strictly within the latent space (avoiding expensive pixel-space decoding). It bridges the giant capacity gap by aligning multi-granularity supervisionโranging from microscopic intermediate features to macroscopic diffusion trajectoriesโwhile dynamically balancing training via a gradient norm adaptive loss weighting mechanism. - **Optimal Synergistic Balancing**: Systematically explores the mutual constraint and upper bound between compact structure and distillation. By mapping this architecture-distillation synergy frontier, we ensure our 0.22B *Moebius* (student) absorbs the maximum semantic reasoning of *[PixelHacker](https://github.com/hustvl/PixelHacker)* (teacher) without triggering representation saturation.
