Z-Cache: Accelerating Diffusion Transformers via Self-Reflection
cs.CV
Submitted: 2026-11-01
Updated: 2026-11-01
Comments: Accepted at NeurIPS 2026 (poster)
Code: https://github.com/black-forest-labs/flux
License: http://creativecommons.org/licenses/by/4.0/
The gist: Diffusion transformers have become the most powerful models for visual generation, but still suffer from massive computation costs.
Terminology
Abstract
Diffusion transformers have become the most powerful models for visual generation, but still suffer from massive computation costs. To solve this problem, feature caching has been proposed to cache the features of diffusion models in the previous computation steps and then reuse them in the following caching steps, which brings significant acceleration but also degradation in generation quality. To address this problem, this paper proposes Z-cache as a feature caching method that can maintain high-quality generation through self-reflection. Concretely, we observe that the error from feature caching tends to be sharply reduced after each full computation. Based on this observation, Z-Cache is designed to first predict the features in the future caching steps and then perform a full computation. After that, Z-Cache returns to the caching steps and re-predicts them based on the previous and the current computation steps, which brings correction in features. Experiments demonstrate that with Z-Cache, diffusion transformers achieve comparable generation quality to the original model but with faster inference speed, for instance, 6.22X acceleration on FLUX-dev for text-to-image generation. Our code is released on GitHub.
Sources
- $\Delta$-DiT: A Training-Free Acceleration Method Tailored for Diffusion Transformers
- CAT Pruning: Cluster-Aware Token Pruning For Text-to-Image Diffusion Models
- Structural Pruning for Diffusion Models
- HunyuanVideo: A Systematic Framework For Large Video Generative Models
- FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space
- Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inference
- Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model
- A Survey on Cache Methods in Diffusion Models: Toward Efficient Multi-Modal Generation
- Step1X-Edit: A Practical Framework for General Image Editing
- Region-Adaptive Sampling for Diffusion Transformers
- DPM-Solver++: Fast Solver for Guided Sampling of Diffusion Probabilistic Models
- ToMA: Token Merge with Attention for Diffusion Models
- MagCache: Fast Video Generation with Magnitude-Aware Cache
- On Distillation of Guided Diffusion Models
- SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis
- Progressive Distillation for Fast Sampling of Diffusion Models
- FORA: Fast-Forward Caching in Diffusion Transformer Acceleration
- Beyond Fixed Formulas: Data-Driven Linear Predictor for Efficient Diffusion Models
- Accelerating Diffusion Transformers with Gaussian Process Rectified Feature Cache
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs