All posts

July 22, 2026

Why we charge a flat fee instead of a token markup

A percentage markup means your gateway earns more when you waste more. Our incentives point the other way — here's the business logic behind $29 flat.

The default business model for AI gateways is a percentage of your traffic — typically around 5% of every token you route. It's easy to bill and it scales with the customer, which is exactly the problem: a markup gateway earns the most when your spend is at its worst.

We charge a flat subscription — $29, $99, or $499 a month — and route your tokens on your own provider keys, at your providers' exact rates. TokenRouter never touches the token bill.

Incentives you can audit

TokenRouter's whole product is spend control: hard caps, rate limits, model access rules, attribution. Every one of those features reduces the tokens you burn. On a markup model, every one of them would reduce our revenue too — a conflict of interest baked into the pricing.

On a flat fee, our revenue is indifferent to your token volume, so we can build aggressive cost controls without hedging. The savings calculator on our compare page will even tell you when we're not the cheap option — below roughly $580 a month of spend, a 5% fee costs less than $29. At that scale you're buying the caps and the controls, not the routing discount.

BYOK is the other half

Bring-your-own-keys isn't just a security preference — it's what makes zero markup verifiable. Your provider dashboards show every token at list price, billed by the provider, and your TokenRouter invoice is the same flat number every month. There's nowhere for a hidden margin to live.

TokenRouter is the flat-price financial ops gateway for LLMs — one endpoint, 13 providers, budgets with hard caps. Start a 14-day trial.