Stable Diffusion WebUI Forge is lllyasviel’s performance-focused rework of the classic Stable Diffusion WebUI. It keeps the familiar interface that millions of users learned, while rebuilding the internals for much better speed and memory behavior, and it has collected about 13,100 stars on GitHub across its many trending appearances.
Forge starts from the AUTOMATIC1111-style Gradio interface but rewrites how models are loaded, patched and moved between GPU and system memory. The practical result is that large models such as SDXL, and later FLUX-class models, run smoothly on graphics cards with modest VRAM that struggle in the original WebUI, and generation is noticeably faster at the same settings. Forge also folds in popular capabilities that previously required separate extensions, including ControlNet-style conditioning and advanced patching tools from the author’s own research, and it maintains broad compatibility with the checkpoints, LoRAs and workflows the community already uses. Development is driven by the creator of ControlNet and Fooocus, and although the pace of updates slowed after mid-2025, Forge remains one of the most installed local image-generation front ends.
Stable Diffusion WebUI Forge is free and open source under the AGPL-3.0 license. Everything runs locally on your own hardware with no subscription or per-image fees.
Forge is for artists, hobbyists and developers who generate images locally and want the classic WebUI workflow with better performance, especially anyone whose GPU has limited VRAM but who still wants to run modern, large diffusion models.