Inference Pro usage in colab

Inference Endpoints on the Hub
1

Hi,

I am new to the Pro account features and “inference pro”. I wanted to try the supported model: Inference for PROs

But I always get the message: “The model codellama/CodeLlama-70b-Instruct-hf is too large to be loaded automatically (137GB > 10GB). Please use Spaces or Inference Endpoints.”

What did I do wrong? Is there really an advantage having the pro account?

thanks and best regards,
Ingo