Training & Fine-tuning

Best AI tools for train models

LoRA / DreamBooth / full fine-tune toolkits.

7 TOOLS INDEXED · HARDWARE-VERIFIED

[ BEFORE YOU CHOOSE ]

The decision that matters

Choose the model family and a reproducible training recipe before choosing the interface. The same GPU can be comfortable for SD 1.5 and constrained for SDXL or Flux.

Check these constraints

Supported model and training method
VRAM at your target resolution and batch
Captioning, sampling, resume, and backup workflow

Read the LoRA hardware guide →

Compare the catalogue

FILTERS //07 / 07 SHOWN

HARDWARE:Max min VRAM:Compute:

Pricing:Sort:7 TOOLS

AI-Toolkit (Ostris)

Modern training framework — Flux, SDXL, SD3 LoRAs in YAML.

OPEN SOURCE16–24 GB VRAM

VRAM fit16–24 GB

FluxGym

Dead-simple Flux LoRA training in a Gradio UI.

OPEN SOURCE12–20 GB VRAM

VRAM fit12–20 GB

Kohya_ss

The standard SDXL/Flux LoRA training UI.

OPEN SOURCE12–24 GB VRAM

VRAM fit12–24 GB

Transformers

The library every LLM ships against first.

OPEN SOURCE2–24 GB VRAM

VRAM fit2–24 GB

Diffusers

Hugging Face's go-to library for every diffusion model.

OPEN SOURCE4–12 GB VRAM

VRAM fit4–12 GB

sd-scripts (Kohya)

The underlying scripts powering Kohya & most LoRA trainers.

OPEN SOURCE8–12 GB VRAM

VRAM fit8–12 GB

OneTrainer

Modern alternative trainer for SD/SDXL/Flux.

OPEN SOURCE12–24 GB VRAM

VRAM fit12–24 GB