How does concurrency billing work?
You pay per second a concurrent slot is occupied by your app, not per token. The minimum billable burst is one second; idle holds keep your slot warm between requests.
Pricing
Start small. Scale on demand.
Pick a one-time plan for predictable concurrency, then extend your usage allowance whenever you need more headroom. The embedded Stripe payment form keeps payment on this page.
Extend usage
Stage usage capacity for spikes.
Extend your usage allowance for bursts, tests, and temporary extra throughput on top of your plan. Every amount opens the embedded Stripe payment form and updates your account after payment confirms.
Frequently asked
Pricing details
Can I switch plans anytime?
Yes. A one-time plan purchase takes effect immediately. Buying another plan updates your active concurrency tier without creating a recurring subscription.
What about token pricing?
Perchy still supports token-priced models for teams that prefer the traditional shape. Browse the catalog on the Models page.