Why your agent bill explodes before the model even thinks — tiered routing and the cheap-check-first fix — Tiered Routing

Description and use

<p>I once watched a single agent workflow burn through $340 in a single afternoon. Not because it failed—because it succeeded. Twenty-three steps, every one of them routed to a frontier model, and nineteen of those steps were formatting JSON, extracting dates, and classifying support tickets. I was paying Cadillac prices for a job a bicycle could do.</p> <p>That was the day I stopped blaming the model and started looking at the router.</p> <h2> The Bill Arrives Before the Model Thinks </h2> <p>The math behind agent cost explosion is deceptively simple. A frontier model like Claude Opus runs 5

Problem

<p>I once watched a single agent workflow burn through $340 in a single afternoon. Not because it failed—because it succeeded. Twenty-three steps, every one of them routed to a frontier model, and nineteen of those steps were formatting JSON, extracting dates, and classifying support tickets. I was paying Cadillac prices for a job a bicycle could do.</p> <p>That was the day I stopped blaming the model and started looking at the router.</p> <h2> The Bill Arrives Before the Model Thinks </h2> <p>The math behind agent cost explosion is deceptively simple. A frontier model like Claude Opus runs 5

Outcome

<p>I once watched a single agent workflow burn through $340 in a single afternoon. Not because it failed—because it succeeded. Twenty-three steps, every one of them routed to a frontier model, and nineteen of those steps were formatting JSON, extracting dates, and classifying support tickets. I was paying Cadillac prices for a job a bicycle could do.</p> <p>That was the day I stopped blaming the model and started looking at the router.</p> <h2> The Bill Arrives Before the Model Thinks </h2> <p>The math behind agent cost explosion is deceptively simple. A frontier model like Claude Opus runs 5

Evidence

Why your agent bill explodes before the model even thinks — tiered routing and the cheap-check-first fix — Tiered Routing

Source