Kohya's GUI

The standard tool for training your own LoRA. Unglamorous, widely used, and the thing every tutorial eventually points at.

Self-hosted
Active
Allowed adult content
Open source

Website Source

At a glance

Where it runs: Self-hostedYou run it on your own machine. Nothing leaves it unless you send it.
Adult content: AllowedAdult content is permitted within the published rules.
Trains on your content: NoYour content is not used to train models.
Status: ActiveMaintained and working today.
Account needed: No
Licence: Apache-2.0
Cost: Free and open source.
What it keeps: Local only. Your training images stay on your disk.

What it takes to run

Graphics memory: 8 GB VRAM, 16 GB recommended
System memory: 16 GB, 32 GB recommended
Disk: about 30 GB
Runs on: Linux, Windows

Find what fits your machine on the tools page — pick your hardware from the first menu.

If you have ever downloaded a LoRA, there is a good chance it was trained with Kohya's scripts. The GUI wraps them in a form, which turns an intimidating pile of command-line flags into an intimidating pile of labelled fields — an improvement, if a modest one.

Training is where the hardware bill lands: a character LoRA on SDXL wants 12 GB and patience.

Worth knowing before you start

Hyperparameters are on you The defaults are a starting point, not a recipe; most first attempts overfit.
Heavy VRAM requirement SDXL training is uncomfortable below 12 GB.

What it does

LoRA and LyCORIS training
Dreambooth
Textual inversion
Dataset tagging helpers

How it relates to other tools

an alternative to OneTrainer

Signs of life

Checked automatically. These are the only figures on this page a machine wrote, and they say when they were taken.

Last answered: Yes, 3 hours ago Its website responded when we asked.
Stars on GitHub: 12,581 1,606 forks.
Last commit: 46 days ago2026-08-01 05:43 UTC
Latest release: v26.0.0
Archived copy: Wayback Machine A snapshot, in case this one stops answering.