DI
SOFTWARE

DiffusionBee

DiffusionBee is a macOS Stable Diffusion desktop interface for local text-to-image, image-to-image, inpainting, ControlNet, LoRA and generation history.

Version 2.5.3macOS 11+ Apple 芯片macOS 12.3.1+ IntelAGPL-3.0; models and outputs may have separate terms

What it is for

DiffusionBee packages text-to-image, image-to-image, inpainting, outpainting, ControlNet, LoRA, upscaling and history into a Mac desktop interface. It is useful for trying local generation without assembling a Python environment. Users who need current models, node workflows or large batch automation may outgrow its slower update pace.

Platform and maintenance boundaries

The current installation packages target Apple Silicon and Intel Mac configurations with different minimum system versions. Windows availability should be treated as a waitlist rather than a released desktop build. Version 2.5.3 was released in 2024, so newer models and workflows may not be supported.

Local generation and model terms

Local inference can keep image content on the Mac, while downloading models and using upload or external features still creates network paths. The application license does not automatically cover Stable Diffusion checkpoints, LoRA files or generated outputs; review each model and asset condition before commercial use.

Maintenance note

This page was reviewed on 2026-08-23 against the 2.5.3 package metadata, system requirements, feature list and AGPL-3.0 note. Start with a small model and copied image before committing to a long workflow.

SAVE TO CLOUD

Save to your cloud drive

Open the cloud drive to get the file directly, or save it for convenient access on another device.

Links checked 2026-08-06
Save first, access when you need itOn desktop, scan with the matching cloud-drive app. On mobile, tap the save button.
GUIDE

DiffusionBee installation and first local generation

Install DiffusionBee 2.5.3 on a supported Mac, prepare one model and generate a small test image with reproducible settings.

Before you start

  • Choose the Apple Silicon or Intel package and confirm the required macOS version.
  • Reserve space for the application, models, history and generated images; large checkpoints can use several gigabytes.
  • Use copied or non-sensitive test images and keep any unique original outside the application history.
01

Installation steps

  1. 01

    Check the Mac architecture

    Confirm Apple Silicon or Intel in System Information, then select the matching package. A wrong architecture can prevent startup or reduce performance.

  2. 02

    Install and open once

    Move the application to the normal Applications location and start it. Review the system warning and file signature before approving the launch.

  3. 03

    Download or import a model

    Use a model supported by the application and wait for a complete download. Check the license, format and source before importing a third-party checkpoint.

02

Quick start

  1. 01

    Generate with a small default setup

    Enter a short prompt, choose one image and a modest size with default parameters to verify the model, history and output folder.

  2. 02

    Change one setting at a time

    Keep the model and seed fixed while adjusting the prompt, steps or guidance scale so the difference is explainable.

  3. 03

    Export and record metadata

    Inspect edges, text, faces and authorization scope, then save the result with the model and parameter notes needed to reproduce it.

Usage tips

  • Intel Macs may run the application but can differ substantially in speed and memory headroom from Apple Silicon.
  • Generated details are model inferences; inspect text, hands, faces and identities before any public or commercial use.
  • A slow release cadence means a new checkpoint may require a newer tool or a different workflow.
Troubleshooting and uninstall

Why is a model download stuck?

Check network access, proxy settings, free space and system time. Keep incomplete files until the application confirms whether a resume operation is available.

Why does generation close the application?

Reduce resolution, batch size and model size, close other memory-heavy applications and test again. Older or low-memory Macs should start with a small workflow.

  1. Export history and model notesSave required images, parameters and the model list outside the application history before removing the app.
  2. Remove the app and optional model dataDelete the application through the normal macOS workflow, then clear models and caches only after confirming that no other task needs them.
FAQ

Frequently asked questions

Is there a released Windows version?

The current package set is organized around Apple Silicon and Intel macOS builds. Treat Windows listings as a waitlist or third-party package until verified.

Are generated images uploaded automatically?

Local generation can run on the Mac, but model downloads and any user-initiated upload or external feature create network activity. Check the actual workflow.

What licenses apply to a generated image?

Check the application license, the checkpoint or LoRA terms and any source asset rights together. The desktop application license does not settle model or output rights.

Which Mac should be used for a first test?

Use a supported system with enough storage and memory, begin with a small model and low resolution, and measure performance before downloading a large collection.