Responses

Endpoint

POST /v1/responses

Authenticate with Authorization: Bearer $OURTOKEN_API_KEY. The endpoint uses the OpenAI Responses input and output shapes rather than Chat Completions messages and choices.

curl https://api.ourtoken.dev/v1/responses \
  -H "Authorization: Bearer $OURTOKEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-5",
    "input": "Explain wallet settlement in one sentence."
  }'

OurToken returns a native response object whose output can contain an assistant message and one or more function_call items. It keeps the public model ID and measured token usage while keeping private provider routing and credentials out of the response.

Function tools and streaming

Function tools use the Responses shape (type, name, and parameters). Return a tool result in a later request as a function_call_output input item with the matching call_id.

Set stream: true to receive native Responses server-sent events, including response.output_text.delta, response.function_call_arguments.delta, and response.completed. The final event follows successful measured-usage settlement; a cancelled request aborts upstream work and releases its hold.

Supported subset

The initial compatibility surface supports stateless text/message input, developer instructions, function tools, function-call output, streaming, and usage. It does not yet support previous_response_id, response storage, background mode, hosted tools, or other provider-specific built-in tools.