LTX Video 13B
Frontierby Lightricks
Highest quality LTX Video model at 13B parameters. Available in dev (best quality), distilled (faster), and FP8 (lower VRAM) variants. Produces high-fidelity video with strong temporal coherence.
- 13B parameters — highest quality in the LTX family
- Dev, distilled, and FP8 variants available
- Requires 32GB+ VRAM at FP16, ~16GB with FP8
- Frames follow 8n+1 pattern (e.g., 81, 161, 257)
- Spatial and temporal upscalers available
Your hardware
Detecting...
Image Quality Benchmarks
Measured quality metrics for LTX Video 13B outputs.
How often humans prefer this model's output (0-100%)
Visual quality and composition rating (5-9 scale)
VRAM by Scenario
VRAM estimates at FP16 and FP8 precision. FP8 uses ~40% less memory with minimal quality loss. Grade shows how well each GPU handles the generation workload.
FP16 (full precision)
| Scenario | VRAM | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 · 25 frames | 25.6 GB | B● | F● | F● | F● |
| 768×512 · 25 frames | 24.0 GB | B● | F● | F● | F● |
| 768×512 · 100 frames | 24.0 GB | B● | F● | F● | F● |
| 1280×720 · 25 frames | 40.0 GB | F● | F● | F● | F● |
FP8 (quantized — ~40% less VRAM)
| Scenario | VRAM | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 · 25 frames | 23.5 GB | B | F | F | D |
| 768×512 · 25 frames | 25.6 GB | B | F | F | F |
| 768×512 · 100 frames | 31.9 GB | D | F | F | F |
| 1280×720 · 25 frames | 34.1 GB | F | F | F | F |
Optimization Tips
Turbo / LCM distillation
Use distilled scheduler at 4-8 steps for faster iteration
Run with Python
from diffusers import LTXPipeline
import torch
pipe = LTXPipeline.from_pretrained(
"Lightricks/LTX-Video",
torch_dtype=torch.float16
)
pipe.to("cuda")
frames = pipe(
prompt="your prompt here",
num_inference_steps=30,
guidance_scale=3.0,
num_frames=257,
).frames[0]
# Save frames or export as videoGet started
Setup instructions for running LTX Video 13B locally
1. Download the model
Get the checkpoint from HuggingFace
2. Place in:
ComfyUI/models/checkpoints/3. Launch ComfyUI
python main.pyMemory Breakdown
VRAM allocation for 25 frames at 768×512 on RTX 4090 24GB
Estimated Generation Time
25 frames at 768×512, 30 steps, FP16.
Sample Outputs
Available Formats & Downloads
Download LTX Video 13B in different precisions. Lower precision = less VRAM but slight quality loss.
| Format | Präzision | Größe | Anbieter | |
|---|---|---|---|---|
| Offizielle Gewichte | ||||
| safetensorsEmpfohlen | FP16 | 26.0 GB | official | Herunterladen |
| Offiziell quantisiert | ||||
| safetensorsOffizielles FP8 | FP8 | 13.0 GB | official-fp8 | Herunterladen |
LoRA Ecosystem
LimitedLTX Video 13B ecosystem is in early stages.
Related Workflows
You might also like
Frequently asked questions
FAQ — LTX Video 13B
How much VRAM does LTX Video 13B need for video?
LTX Video 13B (13B parameters) requires approximately 24.0 GB of VRAM at FP16 precision for generating 25 frames at 768×512. Video generation typically requires more VRAM than image generation due to temporal attention layers.
Can I run LTX Video 13B on RTX 4090?
LTX Video 13B can run on the RTX 4090 with sequential offloading, though video generation will be significantly slower than native fit.
How long does it take to generate a video with LTX Video 13B?
On a reference GPU (RTX 4090 24GB), LTX Video 13B generates a 25-frame video at 768×512 in approximately ~3m 3s at FP16 with 30 inference steps. Faster GPUs with higher memory bandwidth will reduce generation time.
What resolution and frame count does LTX Video 13B support?
LTX Video 13B supports up to 1280×720 resolution and 257 frames per generation at 24 FPS. Higher resolutions and frame counts require proportionally more VRAM.
About LTX Video 13B
See also