<h1 id="ltx2.3-fp4-models"><strong>LTX2.3 FP4 Models</strong></h1><h3 id="because-no-one-uploads-fp4-here.-and-they-should-(fp4-works-on-3xxx-and-above-for-vram-savings.)">Because no one uploads FP4 here. and they should (fp4 works on 3xxx and above for vram savings.)</h3><p>LTX2.3 Distilled FP4ME - Distilled FP4 Mixed Extreme - 14.1GB**<br />LTX2.3 Full Dev FP4ME - Full Dev FP4 Mixed Extreme - 14.1GB**<br />LTX2.3 Official NVFP4 - From Lightricks - Updated 3/17/2026 - 21GB<br />LTX2.3 Dev NVFP4 - from <a target="_blank" rel="ugc" href="https://huggingface.co/Hippotes">Hippotes / </a><a target="_blank" rel="ugc" href="https://huggingface.co/Hippotes/LTX-2.3-various-formats">LTX-2.3-various-formats </a>- 18.15GB<br />LTX2.3 Dev FP4 STFO - from Kijai's Transformers only Scaled FP8 - 16.63GB**<br />LTX2.3 Distilled FP4 - My first FP4 - Transformers Only - 18.15GB**<br /><br />** Are transformers only, or I broke vae/text projection :) needs seperate Vae and Text projection downloads.</p><p>I have a fix for lora issue on FP4ME models. Comfy Bathroom, custom lora loader with presets as well as custom config. <a target="_blank" rel="ugc" href="https://huggingface.co/MrReclusive/LTX-2.3-FP4">https://huggingface.co/MrReclusive/LTX-2.3-FP4</a></p><p>Diffusion model loader/unet loader seems to get confused with these, use the checkpoint loader even though it has no clip/vae</p>

LTX2.3 FP4

<p>My own FP4 conversion from the fp8 scaled model.</p>

ltx23FP4_ltx23DevFP4STFO.safetensors

LTX2.3 Dev FP4 STFO

LTX2.3 Official Dev NVFP4

LTX2.3 Full Dev FP4ME

LTX2.3 Dist FP4

LTX2.3 Distilled FP4ME

LTX2.3 Audio Vae

LTX2.3 Video Vae

LTX2.3 Text Projection

LTX2.3 Dev NVFP4

Files with SHA256: c77d41fbbec99ad6b31503c2c1298e3e900319b16bb473e91c099351cbe8ec20