Models
New sessions start on Auto. Here is how to pick another model, which models you can see depending on how you signed in, and how to set the thinking level.
Auto by default
New sessions use `auto`: the RodiumAi gateway picks a suitable model for each request, so you do not have to choose. Auto is the right choice to start with.
Pick a model
Type /model in the CLI, the desktop app or the editor extension to open the model picker. Models are grouped by vendor, and your choice is remembered. For a single run from the CLI:
…Which models you see
| You signed in with | The list shows |
|---|---|
| RodiumAi account | The models RodiumAi offers for Code, with Auto on top. This list is curated, so it can be shorter than the full API catalogue. |
| API key (BYOK) | Every model your key is allowed to use, from the model catalogue. |
List looks out of date?
Thinking level
On models that support it, the thinking level controls how much the model reasons before it answers. From the CLI, --thinking accepts none, low, medium, high and xhigh. A higher level can give better results on hard problems and spends more of your allowance.
Models and your allowance
With your account, usage is drawn from your FRODI allowance whichever model you pick. With an API key, it is billed in RODI at the rates of the pricing guide. See RODI and FRODI credits.