The list is limited to specific models. This means that custom endpoints that don't use OpenAI models (like Together AI) or selecting specific versions of non-Turbo GPT-4 models is not an available option.
Additionally, it would be nice to have different saved model names call different API endpoints and API keys, for example, having GPT models call the OpenAI API and having Mixtral models call the Together API.
The list is limited to specific models. This means that custom endpoints that don't use OpenAI models (like Together AI) or selecting specific versions of non-Turbo GPT-4 models is not an available option.