Uses / Frontier power, only when it's worth it

Frontier power, only when it's worth it

The privacy objection to local AI is really a power objection: "am I stuck with small models?" Routing answers it.

Day to day, your questions run on your device: free, private, instant. When a question genuinely needs more - current information, hard reasoning, a specialist - routing can escalate exactly that question to a frontier model, and the receipt under the reply tells you which model answered and why.

The modes are the consent: "Auto - Offline Only" never goes online; "Auto - Online and Offline" may, by rules you can read in plain language in Settings. Online use comes out of your plan's monthly allowance with fair metered pricing above it - most days cost nothing.

Online models with prices shown up front, per million tokens

How it works

  1. 1

    Pick a mode per AI

    Offline Only, Online and Offline, or a specific model when you want no routing at all.

  2. 2

    Read the rules

    Settings → Routing explains what happens automatically - health stays local, current-events questions can use web search with cited sources, hard questions can escalate - plus a live list of recent routing decisions and why.

  3. 3

    Tune it

    Prefer fastest or strongest on-device. Privacy-first to freshness-first online. Per-category picks for which online model handles what, with prices shown up front.

  4. 4

    Second opinions

    Under any reply: "Try this answer online" or "Redo on your device". The choice is per-question, never silent.

Download free

Free · No account needed · macOS, Windows, Linux