Can the RTX 3090 24 GB run LTX-Video 13B (0.9.8)?
Yes. Q8_0 fits on the RTX 3090 24 GB with 5.7 GB to spare.
Take Q8_0 (14.0 GB). In use it comes to about 18.3 GB, with 5.7 GB left over on 24 GB.
01What to download
| Part | File | Folder | Size | |
|---|---|---|---|---|
| Model | LTXV-13B-0.9.8-dev-Q8_0.gguf Q8_0 · QuantStack/LTXV-13B-0.9.8-dev-GGUF | models/unet | 14.0 GB | Download → |
| Text encoder | t5xxl_fp8_e4m3fn_scaled.safetensors T5-XXL FP8 (scaled) · comfyanonymous/flux_text_encoders | models/text_encoders | 5.2 GB | Download → |
| VAE | LTXV-13B-0.9.8-dev-VAE.safetensors LTXV 13B 0.9.8 VAE (standalone, QuantStack) · QuantStack/LTXV-13B-0.9.8-dev-GGUF | models/vae | 2.5 GB | Download → |
| Total download · keep about the same free on disk | 21.7 GB | |||
Alternatives: T5-XXL FP16 (9.8 GB). Sizes from Hugging Face, 2026-10-01.
02Every LTX-Video 13B file on 24 GB
| File | Size | Needed | On this card | Quality | Download |
|---|---|---|---|---|---|
| 16-bit SAFETENSORS · Lightricks | 28.6 GB | 32.9 GB | 8.9 GB too big | the original weights | Hugging Face → |
| Q8_0 ← GGUF · QuantStack | 14.0 GB | 18.3 GB | fits · 5.7 GB spare | practically identical to the original | Hugging Face → |
| FP8 SAFETENSORS · Lightricks | 15.7 GB | 20.0 GB | fits · 4.0 GB spare | practically identical to the original | Hugging Face → |
| Q6_K GGUF · QuantStack | 10.9 GB | 15.2 GB | fits · 8.8 GB spare | very close to the original | Hugging Face → |
| Q5_K_M GGUF · QuantStack | 9.8 GB | 14.1 GB | fits · 9.9 GB spare | close; small differences in fine detail | Hugging Face → |
| Q4_K_M GGUF · QuantStack | 8.8 GB | 13.1 GB | fits · 10.9 GB spare | good; some loss in fine detail and text | Hugging Face → |
| Q3_K_M GGUF · QuantStack | 6.5 GB | 10.8 GB | fits · 13.2 GB spare | noticeable loss of detail | Hugging Face → |
| Q2_K GGUF · QuantStack | 4.7 GB | 9.0 GB | fits · 15.0 GB spare | heavy loss; a last resort | Hugging Face → |
“Needed” adds 3.5 GB of working memory for LTX-Video 13B and a 0.8 GB reserve. Sizes 2026-10-01.
03Measured and reported results
Nobody has sent measured numbers for this pair yet. If you run LTX-Video 13B on a RTX 3090 24 GB, send your time per image and peak VRAM and it will appear here, credited.
04About the RTX 3090 24 GB
The RTX 3090 24 GB (Ampere) has no FP8 compute. FP8 files still load and save memory, but ComfyUI converts them back before the maths, so there is no speed gain - Q8_0 GGUF is the closer-to-original 8-bit choice. ComfyUI can compute INT8 files natively on NVIDIA, which is worth a try where a model offers one.
24 GB GDDR6X · 384-bit · 936 GB/s · Ampere. Everything that runs on it →
05Questions
How much VRAM does LTX-Video 13B need?
Around 9.0 GB with the smallest file (Q2_K) and 20.0 GB with an 8-bit file (FP8). The full 16-bit file needs about 32.9 GB.
Which LTX-Video 13B file should I download for the RTX 3090 24 GB?
Q8_0 (14.0 GB) from QuantStack/LTXV-13B-0.9.8-dev-GGUF. Load it with Unet Loader (GGUF) from the ComfyUI-GGUF node pack.
FP8 or Q8_0 for LTX-Video 13B on the RTX 3090 24 GB?
Q8_0 (14.0 GB). The RTX 3090 24 GB has no FP8 maths, so FP8 (15.7 GB) only saves memory and is a step further from the original.
How much do I need to download for LTX-Video 13B on the RTX 3090 24 GB?
21.7 GB: LTXV-13B-0.9.8-dev-Q8_0.gguf, t5xxl_fp8_e4m3fn_scaled.safetensors, LTXV-13B-0.9.8-dev-VAE.safetensors.