Choose an enabled model in Models, create a SingleAPI key in API Keys, and send a chat request from your application. The quickstart shows the complete first request with the OpenAI Python SDK or HTTP. Afterward, use Usage to review totals and charges, or the usage API for an individual request.
| Who | What they do |
|---|---|
| You | Choose the model, set key access and budget, and decide whether to use SingleAPI credits or your own provider key |
| Your application | Sends requests with your SingleAPI key and displays the model's response |
| SingleAPI | Checks access, sends the request to the selected model provider, and records usage |
| Model provider | Runs the model |
Two ways to pay
With managed billing, add SingleAPI credits. SingleAPI uses its provider key and charges your wallet for a completed, billable request. See Credits and usage.
With Bring your own key (BYOK), store a provider key and explicitly select it for a request. Your provider account pays the provider. SingleAPI makes no wallet reservation, deduction, charge or ledger entry for that request. You still need a SingleAPI key to authenticate. See Bring your own key.
What you can send
SingleAPI accepts text chat messages, streaming, JSON response formats on supported models, functions run by your application, provider web search, and Anthropic web fetch on explicitly enabled models. Usage shows traffic metrics; Prompt logs shows captured content when logging is enabled on a key. Model availability, limits and prices come from the catalogue. For exact request fields and unsupported features, use the API reference.