Changing models

The model your assistant thinks with isn't fixed at deploy time — you can

change it whenever you like from the dashboard.

Where

Open your instance and go to the Providers tab. It lists the provider

credentials on this instance and the model currently in use.

Changing the model

Pick a different model from the model list and save — no need to re-enter

your key. The list you see depends on your credential:

  • Before a key is added you see a curated list of that provider's

common models.

  • With your key added the list shows the models your account can

actually use — so if a model is missing, your provider account doesn't

have access to it yet.

Switching providers

Add the new provider's credential on the same tab (see the

provider guides for where each key comes

from), then select one of its models. You can keep several providers'

credentials on the instance and move between them freely.

Choosing sensibly

  • Bigger isn't always better. Flagship models cost more per message

and are often slower; a mid-tier model handles everyday assistant work

well. Step up when you notice quality gaps, not before.

  • Subscription credentials (ChatGPT, Claude) draw on your plan's

limits — heavy use can hit your plan's cap rather than a bill.

  • If a model errors right after switching, check your provider account

can use it (some need billing enabled or access granted), or switch back

while you sort it out.