Why the model "locked" after your first message Guide

When your model picker is set to Auto, the assistant chooses a model based on your first message and then keeps that model for the rest of the conversation. This article explains why it works that way and how to take manual control whenever you want.

What "Auto" does

Auto is the default setting in the model picker. When you send your opening message in a new conversation, Auto reads what you're asking for and selects a model suited to it — a more capable model for complex, multi-step work, or a faster model for quick, simple replies.

Once that first choice is made, Auto stays with the same model for the remainder of the conversation. That's why the picker may appear to "lock" after your first message: the assistant has settled on a model and is keeping it consistent so your replies stay predictable.

Tip: If you want a different model for a task, decide before you send your first message — or simply start a new conversation and pick the model you want up front.

Why it stays fixed for the conversation

Keeping one model for a single conversation gives you a steady, consistent experience. The assistant carries the full thread as context, so holding to one model means the tone, reasoning style, and formatting of its replies don't shift partway through your work.

This is especially helpful for longer tasks — drafting a document across several turns, working through an analysis, or refining a deliverable — where a mid-conversation change in model could produce a noticeably different result.

How to choose a model yourself

You're never limited to Auto. To pick a specific model:

1. Open the model picker near the message composer.
2. Choose a Frontier model for the most capable results on hard or complex work, or a Fast model for quick, straightforward replies.
3. Read the short description shown for each option to help you decide.
4. Send your message as usual.

When you select a model manually, that's the model the conversation uses.

Frontier vs. Fast

• Frontier models are the most capable and are a good fit for demanding tasks such as detailed writing, careful analysis, or multi-step reasoning.
• Fast models prioritize speed and work well for short answers, quick lookups, and simple requests.

Changing the model for a new conversation

Because a model stays fixed once a conversation begins, the simplest way to switch is to start fresh:

1. Start a new conversation.
2. Set the model picker to Auto or to the specific model you want.
3. Send your first message — that sets the model for the new thread.

Your existing conversations remain available and unchanged, so you can keep several going, each with its own model.

Does model choice change anything else?

Different models are built for different strengths, so you may notice differences in how quickly replies arrive and how they read. Frontier models tend to be more thorough on complex work, while Fast models return simple answers more quickly. Choose based on the task in front of you: leave it on Auto to let the assistant decide, or pick a model directly when you already know what a task needs.