Creates a model response using the same interface as the
OpenAI Create a model response
API. Provide a text input string to generate text or JSON outputs. Supports
non-streaming JSON and Server-Sent Events when stream=true.

Omit tools for a simple text response. Send optional tools for function calling (client vs server tools —
see overview above).

Request – Requires model and input. Additional parameters are provider-dependent — see
provider documentation for your model prefix (links in overview).

Tool calling and prompt/context caching are supported when the selected provider and model expose
those capabilities (check model info for supports_function_calling and supports_prompt_caching).

Tools – Client tools return function_call items in output for your application to execute.
Server tools registered on the gateway
run automatically without returning intermediate function_call items for those tools.

Streaming – text/event-stream with data: lines carrying typed events (response.created,
response.output_text.delta, response.completed, etc.). Stream ends with data: [DONE].