Standard or Fast: how Tirrus picks a model

Every AI model is a trade-off between depth and speed. The deeper models reason more carefully and cost more to run. The faster ones answer almost instantly and are perfectly good at most of what people ask in a day.

A long menu of models pushes that trade-off onto you, every time you start a conversation. Tirrus makes it once, carefully, and gives you two clear options.

Fast, most of the time

Fast is the default. Rewriting an email, summarising notes, answering a quick question, drafting a first version: Fast does these well, and it does them without making you wait.

Standard, when it's hard

Standard is for the questions that deserve more care: a contract you're about to sign, a stack trace that makes no sense, numbers that don't add up. It's also the model that reads images, so screenshots, diagrams and photos of the whiteboard go to Standard.

Choosing yourself

You can switch models for any conversation from the model menu. A good rule: if a wrong answer would cost you more than a minute of your time, use Standard.

• Use Fast for writing, rewriting, summarising and everyday questions.
• Use Standard for reasoning, analysis, code and anything with an image in it.

Both are metered the same way

Whichever model answers, the request is metered against your credits like any other, and you can see the usage per request. Standard costs more per answer, which is exactly why Tirrus doesn't use it when Fast will do.