Choosing the AI model (Auto, Frontier & Fast) Guide
How the model picker works in the AI Assistant, what Auto does, and when to choose a Frontier or Fast model.
Last updated July 16, 2026
On this page
The AI Assistant lets you choose which underlying model handles your conversation. You can leave it on Auto and let Praxivara decide, or pick a specific Frontier or Fast model to match the work in front of you.
Where to find the model picker
The model picker sits near the message composer in any chat. Open it to see the available options, each with a short description of what it's best suited for. Your selection applies to the conversation you're in, so different chats can use different models.
What Auto does
Auto is the default. When you send your first message in a conversation, Auto reads the request and selects the model best suited to it — reaching for a more capable model when the task looks complex, and a quicker one when the request is simple.
Once that first message is sent, the model stays fixed for the rest of the conversation. This keeps replies consistent in tone and quality as the chat continues, rather than shifting mid-thread.
Tip: If you know a conversation will be demanding from the start — a detailed proposal, careful analysis, multi-step work — set the model yourself before your first message instead of relying on Auto to infer it.
Frontier vs. Fast
When you'd rather choose directly, you have two families to pick from:
- Frontier — the most capable models. Best for hard, nuanced, or multi-step work where quality and reasoning matter most.
- Fast — optimized for speed. Best for quick, straightforward replies where you want an answer without waiting.
When to choose Frontier
- Drafting long-form or high-stakes content, such as proposals, reports, or client-facing documents
- Analyzing an attached spreadsheet, PDF, or dataset in depth
- Working through multi-step tasks that involve several tools or careful judgment
- Anything where accuracy and thoroughness outweigh a few extra seconds
When to choose Fast
- Quick questions and short answers
- Light rewrites, reformatting, or summaries
- Simple lookups or brief back-and-forth
- Moments when a near-instant reply matters more than depth
How to set the model
- Open the conversation you want to work in, or start a new one.
- Open the model picker near the composer.
- Leave it on Auto to let Praxivara choose, or select a specific Frontier or Fast model.
- Read the short description shown for each option if you're unsure which fits.
- Type your first message and send it. Your choice applies for the rest of that conversation.
If you want a different model, start a new conversation and set it there before sending your first message.
Frequently asked
Why did the model lock after my first message?
Auto and any manual choice both settle in once the conversation begins, so replies stay consistent throughout the thread. To use a different model, start a new chat and choose it up front.
Does the model I pick change what the assistant can do?
No. Every model works with the same assistant features — creating documents and images, searching the web, managing your tasks and contacts, and running your connected apps. The choice affects how the model approaches your request, not which capabilities are available.
Which should I use if I'm not sure?
Leave it on Auto. It's designed to pick a sensible model for most requests, and you can always switch to Frontier or Fast in a new conversation when you have a clear preference.