Pricing

You add credits to your Router account. Credits are deducted per-request based on the model provider's token rate plus the Router fee for your current tier. Your tier is determined by your cumulative monthly spend. No invoices, no surprise charges. If your credits run out, requests pause until you top up.

The first tier covers up to $500/month in inference spend at a 5% Router fee. There is no zero-cost free tier. You pay for the inference itself (to the model provider) plus the Router fee. If you want to test Router before committing, start with a small credit top-up. There is no minimum.

If you've set a budget limit in your dashboard, Router will reject requests once the limit is reached. You'll receive an alert at 80% and 100% of your budget. Requests return a clear error code so your application can handle it gracefully. You can raise the limit at any time from the dashboard.

Tiers are automatic. As your monthly spend crosses a threshold, you move to the next tier and the lower fee rate applies to all subsequent requests that month. You never need to manually switch. At the start of each month, your spend resets and you start at the base tier again.

Router adds a 2–5% fee on top of the provider's listed rate (depending on your tier). In return, you get a single API key, automatic failover across providers, request logging, usage analytics, budget controls, and team management. For most teams, the operational savings from not managing multiple provider accounts and API keys outweigh the Router fee within the first month.

No. The per-token rate you see in our model catalogue matches the provider's published rate. The Router fee is a separate, transparent percentage on your total monthly spend. We do not add hidden margins to token pricing.

Not currently. Router manages the provider relationships and API keys centrally. This is what enables features like automatic failover, unified logging, and single-bill simplicity. BYOK is on the roadmap for the Enterprise tier.