Loras AI

Open-source LoRAs

Loras Library: free open-source LoRA models

Popular community LoRAs for SDXL, Qwen-Image, Z-Image and Wan video, each published under a license that allows commercial use. See the base model, trigger words and license, then get the files from the original author.

SpeedQwen-ImageQwen-Image-LightningDistilled LoRA that lets Qwen-Image generate in 4 or 8 steps.by lightx2vApache-2.0Image editQwen-Image-Edit-2509Qwen-Edit-2509 Multiple AnglesChange the camera angle of an image: rotate, top-down, wide or close-up.by dx8152Apache-2.0VideoWan 2.2 A14BWan2.2-LightningDistilled LoRAs that let Wan 2.2 generate video in 4 steps.by lightx2vApache-2.0VideoWan 2.1 I2V 14BSquish EffectImage-to-video effect that squishes the subject like soft clay.by Remade-AIApache-2.0VideoWan 2.1 I2V 14B360° Rotation EffectImage-to-video effect that spins the subject a full 360°.by Remade-AIApache-2.0SpeedSDXLSDXL-LightningByteDance’s fast LoRA for few-step SDXL image generation.by ByteDanceOpenRAIL++SpeedSDXLLCM-LoRA SDXLLatent consistency LoRA: SDXL images in about 2–8 steps.by latent-consistencyOpenRAIL++StyleSDXLPixel Art XLPixel-art style for SDXL: sprites, scenes and icons.by nerijsCreativeML OpenRAIL-MStyleSDXL3D Render Style XLGlossy 3D render look for SDXL.by goofyaiApache-2.0StyleSDXLLogo.Redmond V2Logo and icon design style for SDXL.by artificialguybrCreativeML OpenRAIL-MStyleSDXLStickers.RedmondDie-cut sticker style for SDXL.by artificialguybrCreativeML OpenRAIL-MStyleSDXLClay Animation.RedmondClaymation look for SDXL.by artificialguybrCreativeML OpenRAIL-MStyleZ-Image TurboClay Animation.Redmond (Z-Image)Claymation look for Z-Image Turbo.by artificialguybrApache-2.0

Every LoRA here was made and published by its original author. We list it with credit and a link to the source, and the license on the source page always applies.

LoRA models: common questions

What is a LoRA model?

A LoRA is a small fine-tuned add-on file for an image or video model. It teaches the base model one style, subject or effect without retraining the whole model, and it only works with the base model family it was trained on.

Are the LoRAs in the Loras Library free for commercial use?

Every LoRA listed here is published under Apache-2.0, OpenRAIL++ or CreativeML OpenRAIL-M, which allow commercial use. Credit the original author, follow any use restrictions in the license, and check the base model’s own license too.

How do I use a LoRA?

Load the LoRA on top of its base model in a tool such as ComfyUI or Diffusers, then add its trigger word to your prompt if it has one. Each LoRA page lists the base model, trigger words and usage notes.

Can I train my own LoRA?

Yes. Upload 4–30 images of one character, product or style on Loras AI and train a LoRA in the cloud, with no GPU or code needed.

Can’t find the right LoRA?

Train your own on 4–30 images of your character, product or style.

Train my LoRA