Serve open text-to-video models on Chalk GPUs
This demo shows how to run LTX-Video’s 13B distilled model on a Chalk GPU to render four-second clips. Describe a shot to try it, then explore the code to see how it works.
Check how a written shot becomes moving video
Start with the rendered examples below: compare each prompt with the subject, motion, and camera in its clip. To try your own, describe a shot and select Generate clip. Each output is about four seconds long; a shared GPU renders one request at a time and shows queue progress.
Watch a rendered example before generating. Compare its prompt with the clip’s motion and camera; Chalk hosts the GPU model behind the application’s render queue.
Built on Chalk
Chalk builds the container and runs LTX-Video on an NVIDIA L40S. The example app manages rendering jobs, reports their progress, and returns playable MP4 clips.
A model ready to use
Deploy the model and its web app together. Chalk provisions the GPU and returns an HTTPS URL.
Deploy from Python
Define the container and its model weights in Python. Later builds can reuse cached image layers.
Call it from your app
Send a text prompt to your model’s HTTPS endpoint and get a video back. Use it in a product feature, a batch job, or an agent.
A render queue in your app
The example server renders one clip at a time and reports each job’s position and progress. Its JSON API lets your application submit a prompt and retrieve the finished MP4.
Run it on your own
Clone the example, log in to Chalk, and run the deploy script. Chalk builds the image, provisions an NVIDIA L40S, and prints the URL for your demo once the model is ready.
You will need uv and a Chalk environment with L40S capacity. The replica downloads about 45 GB of weights at startup. Review the model’s license in the repository before deploying it for others.
git clone https://github.com/chalk-ai/text-to-video
cd text-to-video
# Or export CHALK_CLIENT_ID and CHALK_CLIENT_SECRET.
chalk login
# Builds the image, provisions an L40S, prints the URL of your live demo.
./deploy.py --gpu nvidia-l40sBuild video generation into your product
Turn scene descriptions into clips with a model you deploy. Talk to an engineer about serving video models on Chalk GPUs.