How Forge became the WebUI it is today.
Forge started with the workflow Stable Diffusion WebUI users already knew, then developed its own approach to memory, model loading, image editing, and integrated tools.
Major development milestones
Each phase added capabilities or changed how users set up models and workflows.
- 01
Forge begins
Forge launches as a separate WebUI project with a familiar A1111-style interface and a new backend focused on resource management and development speed.
Read the project update - 02
Memory management becomes a core feature
Never OOM controls, offloading options, and tiled VAE workflows give users more ways to run demanding generations on limited hardware.
Read the project update - 03
Gradio 4 and Forge Canvas arrive
The interface moves to Gradio 4 and gains Forge Canvas for zooming, drawing, masks, panning, and more direct image editing.
Read the project update - 04
FLUX and low-bit loading expand the model stack
Forge adds BitsAndBytes NF4, GPU Weights, swap controls, separate VAE and text-encoder selection, and GGUF-oriented loading routes.
Read the project update - 05
Integrated tools reshape extension workflows
Built-in features and replacement projects change which jobs need a third-party extension and which are available directly in Forge.
Read the project update - 06
Model and workflow support continues to evolve
Later changes cover LoRA behavior, GGUF BF16, Chroma, FLUX sampling, scaled FP8, soft inpainting, and SD Upscale fixes.
Read the project update
Start with the workflow you need
The documentation is organized around tasks rather than release chronology.
Install Forge
Prepare the WebUI for your operating system and GPU.
Install →Add models
Configure checkpoints, VAEs, text encoders, LoRAs, SDXL, and FLUX.
Models →Create images
Move from txt2img to editing, control, upscaling, and advanced workflows.
Workflows →Fix a problem
Diagnose startup, memory, output, model, and extension errors.
Troubleshoot →