Will It Run AI
lightricks

LTX Video 2B

Frontier

by Lightricks

Lightweight 2B video generation model from Lightricks (v0.9.8). Available in dev, distilled, and FP8 variants. The distilled version generates video faster than real-time at lower resolutions. For best quality, use the 13B variant instead.

  • 2B parameters — runs on consumer GPUs (12GB+)
  • Dev, distilled, and FP8 variants available
  • Distilled version: faster-than-real-time generation
  • Frames must follow 8n+1 pattern (e.g., 25, 33, 41, 81, 161)
  • For higher quality, see LTX Video 13B

Your hardware

Detecting...

Parameters2B
Max Resolution1280×720
Max Frames161
FPS24
Architecture3D-DIT
Licenseltx-video-research

Image Quality Benchmarks

Measured quality metrics for LTX Video 2B outputs.

Human Preference Score70%

How often humans prefer this model's output (0-100%)

Aesthetic Score6.5

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)

ScenarioVRAMRTX 4090 24GBRTX 3060 12GBRTX 4060 8GBMacBook Pro M4 Pro 24GB
512×512 · 25 frames13.9 GBSDFA
768×512 · 25 frames12.0 GBSBFS
768×512 · 100 frames12.0 GBSBFS
1280×720 · 25 frames21.7 GBAFFD

FP8 (quantized — ~40% less VRAM)

ScenarioVRAMRTX 4090 24GBRTX 3060 12GBRTX 4060 8GBMacBook Pro M4 Pro 24GB
512×512 · 25 frames12.0 GBSBFS
768×512 · 25 frames14.1 GBSDFA
768×512 · 100 frames20.4 GBAFFD
1280×720 · 25 frames22.5 GBBFFD

Optimization Tips

Turbo / LCM distillation

Use distilled scheduler at 4-8 steps for faster iteration

Run with Python

Run with Python (diffusers)
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=161,
).frames[0]
# Save frames or export as video

Get started

Setup instructions for running LTX Video 2B locally

1. Download the model

Get the checkpoint from HuggingFace

2. Place in:

ComfyUI/models/checkpoints/

3. Launch ComfyUI

python main.py
Note: Video generation requires video output nodes. Install ComfyUI-VideoHelperSuite from the ComfyUI Manager for SaveAnimatedWEBP or VHS_VideoCombine nodes.

Memory Breakdown

VRAM allocation for 25 frames at 768×512 on RTX 4090 24GB

Required: 12.0 GBAvailable: 24.0 GB
Weights4.0 GB
VAE0.2 GB
Text Encoder9.4 GB
Activations6.0 GB
Overhead0.5 GB

Estimated Generation Time

25 frames at 768×512, 30 steps, FP16.

RTX 4090 24GB~1m 28s
RTX 3060 12GB~5m 28s
RTX 4060 8GB~8m 15s
MacBook Pro M4 Pro 24GB~35m 8s

Sample Outputs

Available Formats & Downloads

Download LTX Video 2B in different precisions. Lower precision = less VRAM but slight quality loss.

FormatoPrecisãoTamanhoProvedor
Pesos oficiais
safetensorsRecomendadoFP164.0 GBofficialBaixar
safetensorsBF164.0 GBofficialBaixar
Quantização oficial
safetensorsFP8 oficialFP82.0 GBofficial-fp8Baixar

LoRA Ecosystem

Limited

Very few LoRAs available. LTX Video's ecosystem is still early.

Related Workflows

You might also like

Frequently asked questions

FAQ — LTX Video 2B

How much VRAM does LTX Video 2B need for video?

LTX Video 2B (2B parameters) requires approximately 12.0 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 2B on RTX 4090?

Yes, the RTX 4090 (24 GB VRAM) can run LTX Video 2B at FP16. Expected generation time is around ~1m 28s for a 25-frame clip.

How long does it take to generate a video with LTX Video 2B?

On a reference GPU (RTX 4090 24GB), LTX Video 2B generates a 25-frame video at 768×512 in approximately ~1m 28s at FP16 with 30 inference steps. Faster GPUs with higher memory bandwidth will reduce generation time.

What resolution and frame count does LTX Video 2B support?

LTX Video 2B supports up to 1280×720 resolution and 161 frames per generation at 24 FPS. Higher resolutions and frame counts require proportionally more VRAM.

About LTX Video 2B

Use cases
video-generationtext-to-videoimage-to-video
Recommended runtimes
comfyuidiffusers

See also