Restricting that certain users and groups are only able to access the reasoning and tool use models but restricting the image generation and vison models.

My thought on this is to have guard rails in place that I can give the team access to specific models and LLMs that are focused on their role/need. Basically ring fencing the high cost models that aren’t ideal for day to day conversations, instead of flat out blocking the LLM there is still use cases and usage for it so being able to give access based on a role/group assignment would help control this.