Skip to main content
Protos runs on two LLM tiers, switchable from the composer in the app.

​
Fast

The default tier. Fast, free, and always available — no daily cap, no wallet required. It’s a lighter model, tuned for quick, direct answers: prices, lookups, straightforward swaps.

​
Smart

A more capable model for reasoning-heavy questions — comparing routes, explaining tradeoffs, or answering an open-ended “should I…” question with real context instead of a canned line. Smart has a daily message allowance:

​
What happens at the limit

Hitting the Smart cap doesn’t break anything — Protos degrades gracefully back to Fast rather than refusing to answer. You can also subscribe for uncapped Smart access.

​
Why there’s a cap at all

Smart runs through a paid LLM gateway, unlike Fast. The cap exists so a free tier can exist at all without unlimited inference cost — it’s a cost boundary, not an artificial scarcity gimmick.