Tool 03 / Model Selector

Agent Model Selector: pick by task, stop guessing.

Do you want "more stable code," "cheaper batches," "long-context without losing detail," or "multi-agent scheduling that doesn't fall over"? Pick the task and the priority, and the tool returns a small set of recommendations that match real production work.

What this page does: it does not list every model. It tells you which tier is the most cost-effective starting point for the current task. Before you pay, check the pricing page for the actual rate in your user group.
Selector

What is the task?

Generate the config
Recommendation

Recommended model

Pick a task to see one primary recommendation, two backups, and the reasoning behind the picks.
How to use

Treat recommendations as a starting point, not a rule

  • Get one task fully working on the primary model first, then consider adding the backups.
  • For batch work, look at unit cost and stability before reaching for the most expensive option.
  • Agent-style workflows value "stable across many turns" more than single-turn brilliance.
  • Image generation and code models are separate tracks — don't mix them.
Recommendations shorten the trial-and-error loop; they don't replace validation. Before going live, run at least one real task through the target client.