Your agent's brain: choosing the model

Inside your agent there's a "brain" that thinks up the answers: it's the model. And you get to pick which one to use.

Think of models like car engines: some are compact runabouts (fast and cheap, perfect for everyday errands) and others are sports cars (more powerful for the tough climbs, but they burn more fuel).

/model — see or switch the brain 🟡

It tells you which model this conversation is using, and lets you switch it.

/model
/model Sonnet
/model list

Switch it when: you're about to ask for something big (an analysis, an important piece of writing, code) and you want more power; or the other way around, you want to go faster and cheaper on something simple.

/models — see which engines are available 🟢

Lists the available models, in case you want to choose between them.

/models

Three things worth knowing (so you don't trip up)

• Switching models doesn't erase the conversation. You're only swapping the engine; everything you'd talked about is still there.
• Don't switch it every other message. Each model has its own style; hopping around can change the tone, the consistency, and the cost. Only switch it if there's a clear reason.
• The model you pick by hand is "strict." If that model fails, it'll let you know with an error instead of switching to another one on its own.

Pro shortcut: if you're unsure, just ask it directly "which model do you recommend for this?" and let it decide.

And no, don't fire up the Formula 1 engine to nip out for a loaf of bread: you spend more and you don't get there any sooner. 🏎️🥖