Module gemini_responses 

Direct Gemini GenerateContent ↔ OpenAI Responses buffered conversion.

Modules

stream

Direct, bounded Gemini GenerateContent ↔ Responses incremental conversion. Native source/target collectors validate lifecycle and final canonical parity.

Structs

GeminiReplayContext

The Gemini upstream a Responses request is prepared for, when the caller knows it. Nothing native is restored from it: a signature comes back in the client’s own reasoning items (see signature).

GeminiResponseContext

The original Responses request and effective settings retained during request preparation. Required settings must come from that invocation, not guesses made after receiving a Gemini result. A response conversion consumes this value and recursively removes extensions before reusing same-dialect fields.

GeminiUsageFacts

Functions

gemini_to_responses_request

gemini_to_responses_response

responses_to_gemini_request

responses_to_gemini_response

responses_to_gemini_response_with_modalities