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

Python class

TextGenerationRequestMessage

`TextGenerationRequestMessage`

class max.pipelines.modeling.types.pipeline_variants.TextGenerationRequestMessage(*, role, content='', tool_calls=None, tool_call_id=None, reasoning_content=None)

source

Bases: BaseModel

A single message in a text generation request conversation.

Parameters:

role (Literal['system', 'user', 'assistant', 'tool', 'function'])
content (str | list[TextContentPart | ImageContentPart | VideoContentPart])
tool_calls (list[dict[str, Any]] | None)
tool_call_id (str | None)
reasoning_content (str | None)

`content`

content: str | list[MessageContent]

source

`flatten_content()`

flatten_content()

source

Flattens message content to a role/content dict for text-only messages.

Preserves OpenAI-style tool-calling metadata (tool_calls, tool_call_id, reasoning_content) when set so chat templates that consume conversation history with tool use receive a faithful representation of each turn.

Return type:: dict[str, Any]

`model_config`

model_config: ClassVar[ConfigDict] = {'from_attributes': True, 'frozen': True}

source

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

`number_of_images`

property number_of_images: int

source

Returns the number of ImageContentPart instances in the message content.

`number_of_videos`

property number_of_videos: int

source

Returns the number of VideoContentPart instances in the message content.

`reasoning_content`

reasoning_content: str | None

source

`role`

role: _MessageRole

source

`tool_call_id`

tool_call_id: str | None

source

`tool_calls`

tool_calls: list[dict[str, Any]] | None

source

`validate_content_format()`

classmethod validate_content_format(v)

source

Normalizes message content to a string or list of content parts.

Parameters:: v (Any)
Return type:: str | list[TextContentPart | ImageContentPart | VideoContentPart]

TextGenerationRequestMessage​

content​

flatten_content()​

model_config​

number_of_images​

number_of_videos​

reasoning_content​

role​

tool_call_id​

tool_calls​

validate_content_format()​