[ajithjos]

That's right, we don't provide model access, models are accessed from providers (OpenAI/ Microsoft, Anthropic/Bedrock/Vertext-AI etc).

In the background, yes, its an "api layer" between the models as you said. One advantage is switching models in the API, and the good part is, it comes in with a chatbot infrastructure, like a backend to save chats and a front-end to interact etc.

In simple words, a ChatGPT like app will be readily available for you along with unified API call for models.