D i F low : A System for Micro-Serving Text-to-image Di ffusion Work flows
Open Access
- 28 September 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 533-549
Abstract
Text-to-image generation executes a diffusion workflow comprising multiple models centered on a base diffusion model. Existing serving systems treat each workflow as an opaque monolith, provisioning, placing, and scaling all constituent models together, which obscures internal dataflow, prevents model sharing, and enforces coarse-grained resource management. In this paper, we make a case for micro-serving diffusion workflows with DiFlow, a system that decomposes a workflow into loosely coupled model-execution nodes that can be independently managed and scheduled. By explicitly managing individual model inference, DiFlow unlocks cluster-scale optimizations, including per-model scaling, model sharing, and adaptive model parallelism. Collectively, DiFlow outperforms existing diffusion workflow serving systems, sustaining up to 3× higher request rates and tolerating up to 8× higher burst traffic. We have open-sourced DiFlow at https://github.com/diflow-project/diflow.Keywords
This publication has 23 references indexed in Scilit:
- Helix: Serving Large Language Models over Heterogeneous GPUs and Network via Max-Flow2025
- Medusa: Accelerating Serverless LLM Inference with Materialization2025
- DistriFusion: Distributed Parallel Inference for High-Resolution Diffusion Models2024
- Towards Pareto Optimal Throughput in Small Language Model Serving2024
- Proteus: A High-Throughput Inference-Serving System with Accuracy Scaling2024
- Adding Conditional Control to Text-to-Image Diffusion Models2023
- Tabi: An Efficient Multi-Level Inference System for Large Language Models2023
- INFless: a native serverless system for low-latency, high-throughput inference2022
- Morphling2021
- Atoll2021