Pay for requests, nothing else

There is one plan: prepaid credit, charged per request at the model's listed price. No subscription tier to outgrow, no seats to count, and no negotiation to switch model.

What every account gets

Every model in the catalog, at the listed price
Streaming, vision and image generation
Automatic failover across providers
Per-request usage records with exact charges
As many API keys as you need
No seat count and no monthly minimum

72 models are available at the prices listed in the catalog.

Create an account See model prices

Estimate a month

Live prices

ModelRequests per month10,000Input tokens per request1,000Output tokens per request500

Your monthly spend

$1.58

No subscription. You are charged per request, from prepaid credit.

Questions

How is a request charged?: Credit is reserved when the request starts and settled against actual token usage when it finishes. You pay for what was produced — a request that fails costs nothing, and one you cancel mid-stream costs only the tokens already generated.
Does unused credit expire?: No. Balance stays until you spend it.
What happens when I run out?: Requests are refused with a clear error rather than being queued or silently degraded. Top up and they resume immediately.
Do prices change when you change provider?: No. You are charged our listed price for the model. Which provider serves it — and what they charge us — is our concern, not yours.
Is the API compatible with OpenAI clients?: Yes. Change the base URL and the key; streaming, tool calls and usage reporting behave as your client expects.