At a given time only one model could be selected (no model selected is also possible). If a model is selected, llama-vscode assumes this model is available at the endpoint for this model. If the model is local, the selection of a model starts a llama.cpp server with it.

This way is more clear what models for what will be used.

There are different ways to select a model

• In Llama Agent click the button for selecting a model (completion, chat, embeddings, tools)
• In llama-vscode menu select "Completion models..." (or chat, embeddings, tools)
• Select an env. This will select the models, which are part of the env