# Model Routing

Canonical: https://brew.new/templates/weaveos/model-routing

Brand: weaveos.com
Category: product-update

![Preview of Model Routing](https://cdn.brew.new/email-preview-cWuk3X_FktJ_4uq7D0_CH-1787202241117.png)

## Email content

Weave

PRODUCT UPDATE · PROMPT ROUTING

Every prompt to the right model

Weave now routes each prompt by task complexity. Cheap models handle the boilerplate. Frontier models get the hard reasoning. Your engineers change nothing.

Wooly mascot holding a tablet showing routed prompts

Routing is off by default. Turn it on per workspace.

Enable prompt routing

BEFORE / AFTER · ONE MONTH, ONE TEAM

Single model

100%

of prompts on the frontier tier

Routed

31%

stayed on frontier — the rest moved down a tier

Measured across one design-partner team for a full billing month. Merge rate and review cycles held flat.

Per-engineer AI cost attribution view in WeaveATTRIBUTION

See where the spend actually goes

Token spend is attributed per engineer, per repo, per model. Routing decisions are logged next to the cost they avoided.

See the report →

ROUTING RULES

Works with the tools already in use

Cursor, Claude Code, Devin, and GitHub Copilot route through the same policy. Set a ceiling per team and Weave holds it.

Read the breakdown →

Gold token coins representing model token spend

routing · token spend · model cost

What routing does not change

No prompt content is stored. No per-engineer scoring. The policy sees task shape and cost — apples to apples, every time.

Two Wooly mascots high-fiving over a cheaper bill

Turn routing on today

One toggle per workspace. You can revert it in a click.

Enable prompt routingor book a demo

Weave logo

Engineering intelligence for AI-native teams.

LinkedIn

Weave · San Francisco, CA

Weave © 2026

Unsubscribe

[Open and remix this design](https://brew.new/templates/weaveos/model-routing)

[Browse email designs](https://brew.new/browse/templates)