Investigate how deploy multiples models on demand. We want to get the same feature Ollama has, download and expose multiples models, so the user can specify what model want to use.

This can be achieved using Openshift AI multi-model serving: https://docs.redhat.com/en/documentation/red_hat_openshift_ai_cloud_service/1/html/deploying_models/deploying_models_on_the_multi_model_serving_platform

edit: Multi-modal in Openshift AI 2.19 is deprecated, so this is not a viable option.

We should investigate if there are other alternatives, because maybe we can discard using OpenShift AI