Choosing a Model
AI CTRL gives you access to several AI models at once. This article covers how to pick one, and how to stop thinking about it if you'd rather not.
The short answer
Use the Smart Model Router. It reads what you've asked for and assigns the model best suited to it, choosing only from models you already have access to. For most people doing most things, that's the whole answer, and you can stop reading here.
The rest of this article is for choosing deliberately.
The model list
Click the model name at the top of the chat area to open the list. Everything your organization is licensed for appears here, each with a short note describing what it's good at.
Three things worth setting up once:
- Pin the models you use to keep them in the sidebar, a click away from any screen
- Set your default model in your settings: this is what opens when you start a new chat
- Switch mid-conversation whenever you want; everything discussed so far carries over
Your administrators decide which models appear. New ones are added regularly once they've passed our security testing, but they arrive restricted: if there's a model you want and can't see, ask rather than assuming it isn't available.
The current model list, and what each costs to run, is on the AI CTRL Ratecard.
What kind of model is this?
Matching the type to the task matters far more than picking the exact right name.
| Type | Best for | How you'll recognize it |
|---|---|---|
| Everyday / lightweight | General questions, drafting, summarizing, rewriting, quick lookups | Finishes first in a side-by-side, and sits at the low end of the ratecard. Handles more than people expect. Try it first |
| Reasoning | Multi-step problems, analysis, technical writing, code | Slower, more verbose than a short question needs, and the most expensive per credit |
| Large context | Long documents, large data sets, anything used with the Structured Data Tool | Use when the volume is genuinely large, not by default |
| Specialized | Image generation and other single-purpose tasks | Availability depends on your organization's configuration |
If a model's description doesn't make its purpose clear, ask your administrator. They chose your list, and why.
Finding out which you prefer
You don't have to choose on faith. Send one prompt to several models at once and read the answers side by side; it teaches you more in thirty seconds than any description can.
- Start a new chat and select a model
- Click + beside the model name to add a second
- Add a third if you want. Stop at three: more is hard to read
- Send your prompt once. All lanes answer it
- Click − beside a model name to drop it
What to watch for:
- One finishes noticeably sooner. That's your lightweight model, and for a lot of everyday work it's all you need.
- The tone differs. Read a line from two lanes back to back. Neither is wrong, but you'll have a preference, and now you know what it is.
- Ask something subjective. Factual lookups produce near-identical answers, which is useful in itself: if they all agree, you didn't need the expensive one.
When you've decided, drop the models you don't want: click − until only your preferred lane is left, then carry on from there. Every model still in the comparison answers every message you send, so three lanes left open means paying three times for the whole conversation, not just the prompt you were comparing. Worth it while you're working out what you prefer; not something to leave running all day.