What it is
The model picker lives right in the chat box. It lets you switch which AI model you're talking to on the fly — mid-thread, without starting a new conversation. Your admins pre-select which models are available to choose from, so every option in the picker is one your company has already approved.
Why it's useful
Different models are better at different things. With the model picker, you can match the model to the task without leaving your current thread:
Switch to a faster or cheaper model for quick, structured tasks (summarizing, formatting, simple Q&A).
Switch to a more powerful model for harder reasoning, complex analysis, or nuanced writing.
Compare how different models respond to the same question.
Your conversation history carries over when you switch — you're not starting from scratch, just changing which model handles the next message.
How to use it
Click on the AI Provider icon in the chat box.
Choose from the list of models your admins have enabled.
Continue chatting and your next message (and the rest of the thread) is handled by the newly selected model.
Important: delegation works differently
If your message causes elvex to delegate your task to an existing agent (a specialized assistant built for a specific job), that agent uses its own assigned model which its builder configured, and not the model you currently have selected in the picker.
In other words:
The model picker controls the model for your direct conversation with elvex.
It does not override the model an agent was built with. Each agent's model is set by whoever created it.
If you need a specific model's behavior for a task, keep it in your own thread rather than routing it to an agent — or check with the agent's builder about which model it uses.