For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Python module

max.pipelines.request

Request types for MAX API.

Submodules

max.pipelines.request.provider_options	Provider-specific options for MAX platform and modalities.

Base request types

Request	Protocol representing a generic request within the MAX API.
RequestID	A unique immutable identifier for a request.
RequestType	Type variable for request types.

DUMMY_REQUEST_ID	Stand-in request ID for work that has no client request, such as the warmup pass run while capturing a CUDA graph.

Responses API request

OpenResponsesRequest	General request container for OpenResponses API requests.
OpenResponsesRequestBody	Request body for creating a response.
ResponseResource	The response from the model.

Messages

AssistantMessage	A message from the assistant.
DeveloperMessage	A message from the application developer.
InputMessage	A message the client sends with a request, in any input role.
Message	A generic message in the conversation.
MessageRole	The role of the message author.
MessageStatus	The status of the message.
SystemMessage	A system message that guides the assistant's behavior.
UserMessage	A message from the user.

Input content

InputContent	The content parts an input message may carry: text, an image, a file, or video.
InputFileContent	File content in input messages.
InputImageContent	Image content in input messages.
InputTextContent	Text content in input messages.
InputVideoContent	Video content in input messages.
ItemReferenceParam	Reference to an item in the conversation.
ImageDetail	Controls image processing detail level.

Output content

OutputAudioContent	Audio content generated by the model in output messages.
OutputContent	The content parts an output message may carry, deserialized through the type field of each part.
OutputImageContent	Image content generated by the model in output messages.
OutputTextContent	Text content in output messages.
OutputTokensDetails	Detailed token count information for output.
OutputVideoContent	Video content generated by the model in output messages.
RefusalContent	Refusal content when the model declines to respond.

Tools

AllowedToolChoice	Specifies a subset of allowed tools and selection mode.
FunctionCall	A function call generated by the model.
FunctionCallOutput	The output from a function call.
FunctionCallStatus	The status of the function call.
FunctionTool	A function tool definition in the response.
FunctionToolChoice	Specifies a particular function to call.
FunctionToolParam	A function tool definition in the request.
ToolChoice	How a request constrains tool use: a preset mode, a forced function call, or a subset of allowed tools.
ToolChoiceValueEnum	Controls which (if any) tool is called by the model.

Response format

JsonObjectField	JSON object response format.
JsonObjectParam	Configuration for JSON object responses.
JsonSchemaField	JSON schema response format.
JsonSchemaParam	Configuration for JSON schema-constrained responses.
ResponseFormat	The response format reported back in a response: plain text, a JSON object, or a JSON schema.
ResponseFormatParam	The client-supplied form of a response format: plain text, a JSON object, or a JSON schema.
TextField	Text response format.
TextParam	Configuration for text generation.

Reasoning

ReasoningBody	Reasoning information in the response.
ReasoningEffortEnum	The effort level for extended reasoning.
ReasoningParam	Configuration for extended reasoning.
ReasoningReference	Reference to reasoning in the response.
ReasoningReferenceParam	Reference to previous reasoning content.
ReasoningSummaryContent	Summary of the model's reasoning process.
ReasoningSummaryEnum	The level of detail for reasoning summaries.

Usage and tokens

AudioGenerationDetails	Audio generation usage metadata.
Error	Error information.
ImageGenerationDetails	Image generation usage metadata.
IncludeEnum	Additional data to include in the response.
IncompleteDetails	Details about why a generation was incomplete.
InputTokensDetails	Detailed token count information for input.
LogProb	Log probability information for output tokens.
ServiceTierEnum	The service tier for request processing.
StreamOptionsParam	Options for streaming responses.
TopLogProb	Log probability information for a token.
TruncationEnum	Controls message truncation behavior.
UrlCitationBody	URL citation in response.
UrlCitationParam	URL citation in request.
Usage	Token usage statistics.
VerbosityEnum	The verbosity level for responses.

Protocols

FastAPIRequestProtocol	Minimal protocol for FastAPI/Starlette Request objects.
MediaDataUriFetcher	Fetches a web media URL and returns it as a base64 data: URI.