Skip to main content
Mesa gives you a choice of AI models and effort levels for every conversation. Matching them to the task at hand gets you better results while making the most of your usage allocation.

​
Select a model and effort level

Use the dropdown above the chat box on the home page to pick a model and effort level before you send your message. The picker leads with Auto and the three models that work best in Mesa; the rest are tucked under More models.

​
Compare models

Leave the model on Auto to let Mesa pick the best one for each task, or choose one of the three recommended models: Open More models for additional options: Gemini 4 Argon (Google’s most intelligent model), Gemini 3.8 Flash (a fast Google model), Claude Sonnet 5.5 (a balanced Anthropic model), and Claude Haiku 5.5 (a fast Anthropic model).
The model picker with the More models section expanded
Older model versions have been moved out of the picker to keep the choice simple. A conversation that’s already pinned to an older model keeps using it.
If your organization or team sets a default model, Auto starts new conversations on that model instead. See Organization settings and Teams.

​
Compare effort levels

​
Which combination should you pick?

  • Quick lookup or simple question: a lighter model on Fast. You’ll get an answer in seconds without dipping into your allocation.
  • Drafting a document or investigating a specific issue: a higher-capability model on Balanced.
  • Comprehensive research, many documents, or a polished deliverable: a higher-capability model on Deep Research. Pair it with Notifications so you don’t have to wait at the screen.
Higher-capability models and effort levels consume more of your allocation. See Usage to check how much you have left.

Usage

Track daily and weekly usage so you can plan your work.

Writing effective prompts

A well-written prompt matters as much as the model you pick.

Batches

Run one task across many items when completeness matters more than speed.
Last modified on October 20, 2026