New: MiniMax H3 2K — native 2K video with synchronized audio, first/last frames, and multimodal references. Choose it in the video model menu.

New: Seedream 5.0 Pro — ByteDance's highest-quality Seedream model for text-to-image and multi-reference edits.Try it

New: Gemini Omni Flash video — turn text or images into video with synchronized audio. Choose it in the video model menu.

Models Gallery

Made available by the community

Checkpoint LoRA Textual Inversion

All sd15 sdxl1 flux1

SDXL-Turbo

Animagine XL V3 v3.0

AnyLoRA - Checkpoint bakedVae ftMse fp16 NOT-PRUNED

PonyXL

万象熔炉 | Anything V5/Ink ink

XXMix_9realistic v4.0

OnlyRealistic | 《唯》· 超高清真人写实 v30 Baked VAE

BigAsp

InteriorDesignSuperMix V2

RealCartoon-XL V7

SDXL HK v0.6

Real Dream 14 (SD1.5)

POLYHEDRON SDXL v3

SDXL Turbo Unstable Diffusers ☛ YamerMIX V10 TURBO EDITION

Copax TimeLessXL - SDXL1.0 v12.2

RealVisXL V5.0 V5.0 (BakedVAE)

Counterfeit-V3.0 v3.0

epiCRealism XL V9-UnFlux

Pony Realism 🔮 🌐 v2.2 Main + VAE

RealVisXL V3.0 + Turbo V3.0 (BakedVAE)

Models can be used to create new fine-tunes by using the Advanced - Base tune dropdown in the new fine-tune form.
This feature is in beta
Fine-tunes results will generally be less similar to the original subject
Models are uploaded and published by the community, are subject to review, and may be removed at any time pending license and review.
To use LoRa use syntax <lora:token:multiplier>, e.g: <lora:nijiMecha:0.8> - multiplier should be between 0.1 and 1.0
To use Textual Inversion add the token to the prompt text or negative prompt.