Choose a model for a chat
1
Open the model picker
In the chat box, click the model name.
2
Find a model
Models are grouped by provider, such as OpenAI or xAI. Type in Search models… to filter the list. Each model shows its context size and its price per million output tokens.
3
Pick it
Click the model. The chat uses it from your next message.
How Auto works
Before each message, Auto asks a small, fast model which of three tiers the message needs. It’s told to pick the cheapest tier that will still do the job well.
Auto also follows these rules:
- If it isn’t confident about a tier, it moves up one tier.
- A message with several tasks, or with text files attached, uses at least the standard tier.
- A message that’s likely to change production, such as publishing, deploying, writing through the Fluid API, issuing a refund, writing to a database or deleting something, uses the complex tier.
- In a long chat that was active in the last 5 minutes, it doesn’t drop to a cheaper model than the one it used last. There, a cheaper model would cost more, because it can’t reuse the chat the current model has already cached.
- If it can’t choose within about 2.5 seconds, or the choice fails, it uses Claude Opus 5.5.
Set the default model for new chats
1
Open Preferences
Click your name at the bottom of the sidebar, then click Preferences.
2
Choose the default
Under Default chat model, pick a model, or Auto — Mist recommended.
Frequently asked questions
Does the model I choose change the cost?
Does the model I choose change the cost?
Yes. Each model has its own price, which the picker shows. Your company pays for Mist’s AI usage, up to a daily cap. See your company’s usage under Settings > AI usage in Mist, or on AI Spend in the admin.
Why did my chat switch back to Auto?
Why did my chat switch back to Auto?
The model you picked is no longer available. Mist moves the chat to Auto when that happens.
What happens when my company runs out of AI credits?
What happens when my company runs out of AI credits?
Mist stops answering and tells you why, for example “Your company is out of AI credits.” or “Your company’s AI spend limit is reached.” An admin can check Billing in the admin.