RuneXX

/

LTX-2.3-Workflows

like 268

In case someone is stuck on where to put the files.
The file structure is as follows for the files downloaded :

-- ComfyUI/models
---------diffusion_models
--------------- ltx-2.3-22b-dev/distilled_transformer_only_XXX .safetensors (fp8 and fp16)
--------- text_encoders
-------------- gemma_3_12B_it_XXXX.safetensors (fp8, fp8, scaled etc)
-------------- ltx-2.3_text_projection_bf16.safetensors
--------- vae
-------------- LTX23_audio_vae_bf16.safetensors
-------------- LTX23_video_vae_bf16.safetensors
-------------- taeltx2_3.safetensors (tiny vae for sampler previews, optional)
--------- loras
-------------- ltx-2.3-22b-distilled-lora-XXX.safetensors (currently there are many variants, but the one at Kijai works great)
--------- latent_upscale_models
-------------- ltx-2.3-spatial-upscaler-x2-1.1.safetensors
--------- Unet
-------------- any LTX gguf model

Where to find the models

Most of the models are here :https://huggingface.co/Kijai/LTX2.3_comfy/tree/main
Text encoders are here: https://huggingface.co/Comfy-Org/ltx-2/tree/main/split_files/text_encoders
Upscale models are here: https://huggingface.co/Lightricks/LTX-2.3/tree/main

GGUF models are found many places around Huggingface, for example Quantstack, Unsloth and more
https://huggingface.co/QuantStack/LTX-2.3-GGUF
https://huggingface.co/unsloth/LTX-2.3-GGUF