veloxML: realtime ML deployments

test_endpoint.ipynb

In [ ]:

import requests

response = requests.post(
    "...",
    json={"text": ""}
)
response.json()

0.0s

Out[1]:

bash

What is veloxML?

veloxML allows you to take your model to the cloud with ease, speed and reliability.

In practical terms, any directory that contains a requirements.txt file can be transported to the cloud with one command: veloxml deploy.

Every time you deploy a project, veloxML gives you a unique URL to it (even before build processes are complete!). These URLs look like this: my-model-hj1v2m.veloxml.com.

When it's time to take your deployment to production, you simply pick an appropriate alias.

You can think of veloxML as the Vercel for ML models.

PyTorch

                    $ my-model/ ls

                    model.pt predict.py requirements.txt

                    $ veloxml deploy

HuggingFace

                    $ my-transformer/ ls

                    model.safetensors predict.py requirements.txt

                    $ veloxml deploy

vLLM Native

                    $ my-llm/ ls

                    model.safetensors

                    $ veloxml deploy --engine vllm

Stop fighting infrastructure.

Get instant access to serverless GPU scaling and one-click deployments.

Due to high demand, we are currently onboarding users manually. We will reach out to you shortly.