Choosing the model

Every agent runs on an underlying AI engine, and each engine offers a few different models — think of them as the same agent thinking at different speeds and depths. A bigger model reasons harder and catches more, but takes longer; a lighter one answers fast. Each agent already has a sensible default model picked for it, so you never have to think about this. But when a task is unusually hard, or unusually quick, you can change the model for a conversation yourself.

Switching the model

Look at the small gear button just to the left of the message box — that’s the model menu. Click it and you’ll see the models available for this agent’s engine, with the current one checked. Pick a different one and you’re done; the menu closes and your next message goes to the model you chose.

A couple of things worth knowing:

• The choice belongs to the conversation, not the agent. When you switch models, it sticks to this session and is remembered if you come back to it later. Your other conversations — and the agent’s default everywhere else — stay exactly as they were.
• The list depends on the engine. Different engines offer different models, so the menu only ever shows the ones the agent in front of you can actually run. You won’t see a choice that wouldn’t work.

Picking a good one

You don’t need to know the model names. The rule of thumb is simple:

• Hard, careful, or high-stakes work — a tricky refactor, a long document, a problem that’s bitten you twice — reach for a heavier model. “Use the strongest model for this one” is a perfectly good instinct.
• Quick, simple, or throwaway work — a one-line answer, a small edit, a fast question — a lighter model gets you there sooner.

If you’re not sure, leave it on the default. It’s chosen to be a good all-rounder.

Fast mode

Some models also offer a fast mode — a toggle that appears at the bottom of the model menu, but only when the model you’ve selected supports it. Turn it on and that engine prioritizes getting words on the screen quickly. If you’ve picked a model that doesn’t offer it, the toggle simply isn’t there — nothing to configure, nothing to miss.

Staying in control

Changing the model never changes what the agent is or what it’s allowed to do — only how it thinks. It’s a dial you can turn up or down at any point in a conversation, and turn back just as easily.

What to read next

• Who’s doing the work, and how engines map to agents → /agents/
• How conversations and turns work → /sessions/