Providers — Data Flow

Gemini Text

PromptResult | str
    -> GeminiProvider.generate_text()
    -> google.genai.models.generate_content()
    -> response.text
    -> str

GeminiProvider.answer() delegates to generate_text() for compatibility with LLMDispatcher.process().

Gemini JSON

PromptResult | str
    -> GeminiProvider.generate_json(schema=...)
    -> GenerateContentConfig(response_mime_type="application/json", response_schema=schema)
    -> response.text
    -> json.loads()
    -> optional Pydantic validation

Gemini Images

prompt: str
    -> GeminiProvider.generate_image_bytes()
    -> GenerateContentConfig(response_modalities=[IMAGE], image_config=...)
    -> first inline_data image part
    -> (bytes, actual_mime_type)

With input_images, the provider sends multimodal content:

prompt: str + input_images: list[ImageInput]
    -> text part + inline_data image parts
    -> GeminiProvider.generate_image_bytes()
    -> first returned inline_data image part
    -> (bytes, actual_mime_type)

response_mime_type is not passed to GenerateContentConfig.response_mime_type on this path; it is only a fallback content type when Gemini omits inline_data.mime_type. When image_config.image_size is 4K and Gemini rejects the request, the provider retries once with image_size changed to 2K.

Imagen Images

prompt: str
    -> GeminiProvider.generate_imagen_bytes()
    -> GenerateImagesConfig(output_mime_type=requested_mime)
    -> first generated_images image
    -> (bytes, actual_mime_type)

OpenAI Text

PromptResult | str
    -> OpenAIProvider.generate_text()
    -> _OpenAIRequestMapper
    -> responses.create(store=False)
    -> response.output_text
    -> str

OpenAI Structured JSON

PromptResult | str + Pydantic schema
    -> OpenAIProvider.generate_json()
    -> responses.parse(text_format=schema)
    -> response.output_parsed
    -> BaseModel

Without a schema, the provider requests JSON mode and decodes response.output_text.

OpenAI Streaming

PromptResult | str
    -> OpenAIProvider.stream_text()
    -> responses.create(stream=True)
    -> response.output_text.delta events
    -> AsyncIterator[str]