Making a Request

Making a request is the same as sending a standard API call to a Provider, such as getting a chat completion from Anthropic. Pay-i will automatically calculate the cost of these calls, using the latest prices and accounting for all the tricky nuances, and will augment the response from the AI provider with an xproxy_result object telling you exactly how much was spent in the request and the status of any limits involved in the request.

Pay-i also allows you to send additional, optional parameters which are used to categorize and/or control the costs associated with the API calls. For more information, see the Data Association and Pivots section.

Sending Custom Headers

If you want to get the most out of Pay-i's features for understanding and controlling costs, additional headers can be sent in when making requests. Most Provider SDKs support an “extra_headers” parameter which allows for these headers to be sent in as part of the natural use of their APIs. Pay-i uses these to avoid requiring you to rewrite your code to use a new SDK to make requests.

OpenAI Custom Header Example

response = client.chat.completions.create(
  model=model,
  messages=messages,
  extra_headers=<payi_headers>
)

Anthropic Custom Header Example

message = client.messages.create(
  model=model,
  max_tokens=max_tokens,
  system=system,
  messages=messages,
  extra_headers=<payi_headers>
)

Each feature section describes the custom header(s) required to leverage that Pay-i feature or scenario.

More detailed examples are also available in the Pay-i Quickstarts.

Header Forwarding

If you would like to send additional, arbitrary headers to the model provider, see Header Forwarding.

Related APIs

• OpenAI Requests
• Azure OpenAI Requests
• Anthropic Requests
• Ingest Events