How plans and credits work
Metir runs on credits. Your plan gives you a monthly allowance, and each thing you ask for spends a little of it. Heavier work spends more.
Monthly allowance
An ordinary chat message
The same question on Genius
A Deep Research run
A generated image
Auto picks the cheapest model that can answer
- Your plan is an allowance
- Routine questions are cheap
- Research and generation cost more
- Auto picks the cheapest model that can answer
What spends credits
- An ordinary chat message.
- A message on a stronger model. See choosing a model.
- A Deep Research run, which is the most expensive thing on this list because several agents read across your documents.
- Generating an image. See images.
- Transcribing a recorded meeting.
- A Blob doing work on its own.
What does not spend credits
- Browsing your Workspace.
- Reading a transcript or opening a file.
- Searching your chat history.
Controlling what you spend
- Leave the category chip on Auto and let Metir choose. A simple question then does not cost what a hard one does.
- Use Smart for routine work and save Genius for problems that genuinely need it.
- Scope Deep Research to the folder that matters rather than your whole library. See choosing your perimeter.
Where to see your balance
Your balance and plan live on the subscription page in your account settings. The sidebar only prompts you about your plan when you are running low. Full detail on managing your subscription.
Upgrading
On the web you upgrade in the browser. In the iOS and Android apps the purchase goes through the App Store or Google Play. Whichever route you use, the plan applies to your account everywhere.
Common questions
What happens when I run out of credits?
Metir tells you, and you can upgrade or wait for your allowance to reset. Nothing you have already created is affected.
Does browsing my own files cost anything?
No. Reading, searching and opening things in Workspace is free. Credits are spent when a model does work.
Why did one question cost more than another?
Because of what it asked for. A one-line answer on a fast model is cheap; a research run across a folder of documents involves several agents reading a lot of text.
Pricing and allowances change, so this page describes how spending works rather than listing numbers. Current prices are on the pricing page.