Weave Router Escalation Budget

Treat Weave Router as a two-tier spend policy, not a single setting. The router reads every agent turn, classifies complexity in single-digit milliseconds, and sends routine work to DeepSeek, Llama, or Gemini while holding Claude and GPT for stalled or looping tasks. For an agency, the framework is simple: set a monthly frontier ceiling per client, then let the escalation classifier decide when to cross it. A dev shop on the $1,800 Weave Router Starter Setup can cap frontier spend at, say, 15 percent of total token cost and review the usage dashboard weekly. If escalations climb above that band, the cause is usually prompt quality or task decomposition, not the router. The Pro plan at $50 monthly covers drill-down and team stats for that review. Enterprise pricing is custom and quoted on contact, so multi-client agencies should request terms before standardizing.