We’re giving way free GPUs for a week to test our system. Email us at if you want access.

To deploy a GPU, first, create an account. Then, select the Docker option:

In the Source code section, drop a .zip file with your code. Or use oen of the examples (download them and zip them):

🌔 Image Q&A

FastAPI app that uses moondream2 to answer questions about an uploaded image.

🦙 llama-cpp server

A a web server that allows using Llama 2

🪐 JupyterLab

Connect to a remote Jupyter server to fine-tune LLMs.

Then, in the deployment form, select 1 GPU in the Advanced section:

Currently, only 1 GPU is supported, which will deploy your application on a machine with an NVIDIA T4 (16GB), 4 CPUs and 16 GB of RAM.