The honest limit
On CPU, small models of a few billion parameters run, slowly. Larger models need a GPU, and we do not rent GPU servers. If a model does not fit in RAM, a bigger VPS will not make it fast.
Option one: the Ollama template
App Hosting has an Ollama and Open WebUI template, CPU only and meant for small models. It starts on the Business plan. It is the quickest way to try a model with a chat screen.
Option two: Ollama on your own VPS
- Create the server
Pick Ubuntu 24.04 with enough RAM for the model you plan to load.
- Install Ollama
Run the official installer and pull one small model first.
- Keep the API private
Ollama has no login by default. Bind it to localhost or put it behind a firewall and a reverse proxy with authentication.
- Measure
Test the speed on your real prompts before you build on it.
When to use the API instead
Use the AI API for bigger models, or when you need steady speed without managing a server. It is OpenAI-compatible, with the base URL api.bookmyhost.com/v1, so most clients work by changing the URL and key.


