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

OpenResponsesRequestBody

`OpenResponsesRequestBody`

class max.pipelines.request.OpenResponsesRequestBody(*, model, input, citations=None, frequency_penalty=None, include=None, logit_bias=None, logprobs=None, max_output_tokens=None, metadata=None, modalities=None, n=None, parallel_tool_calls=None, presence_penalty=None, reasoning=None, reasoning_reference=None, response_format=None, seed=None, service_tier=None, stop=None, store=None, stream=None, stream_options=None, temperature=None, tool_choice=None, tools=None, top_logprobs=None, top_p=None, truncation=None, user=None, verbosity=None, provider_options=<factory>)

source

Bases: BaseModel

Request body for creating a response.

Parameters:

model (str)
input (str | list[UserMessage | AssistantMessage | SystemMessage | DeveloperMessage])
citations (list[UrlCitationParam | ItemReferenceParam] | None)
frequency_penalty (float | None)
include (list[IncludeEnum] | None)
logit_bias (dict[str, float] | None)
logprobs (bool | None)
max_output_tokens (int | None)
metadata (dict[str, str] | None)
modalities (list[str] | None)
n (int | None)
parallel_tool_calls (bool | None)
presence_penalty (float | None)
reasoning (ReasoningParam | None)
reasoning_reference (ReasoningReferenceParam | None)
response_format (TextParam | JsonObjectParam | JsonSchemaParam | None)
seed (int | None)
service_tier (ServiceTierEnum | None)
stop (str | list[str] | None)
store (bool | None)
stream (bool | None)
stream_options (StreamOptionsParam | None)
temperature (float | None)
tool_choice (ToolChoiceValueEnum | FunctionToolChoice | AllowedToolChoice | None)
tools (list[FunctionToolParam] | None)
top_logprobs (int | None)
top_p (float | None)
truncation (TruncationEnum | None)
user (str | None)
verbosity (VerbosityEnum | None)
provider_options (ProviderOptions)

`citations`

citations: list[UrlCitationParam | ItemReferenceParam] | None

source

`frequency_penalty`

frequency_penalty: float | None

source

`include`

include: list[IncludeEnum] | None

source

`input`

input: str | list[InputMessage]

source

`logit_bias`

logit_bias: dict[str, float] | None

source

`logprobs`

logprobs: bool | None

source

`max_output_tokens`

max_output_tokens: int | None

source

`metadata`

metadata: dict[str, str] | None

source

`modalities`

modalities: list[str] | None

source

`model`

model: str

source

`model_config`

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

source

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

`n`

n: int | None

source

`parallel_tool_calls`

parallel_tool_calls: bool | None

source

`presence_penalty`

presence_penalty: float | None

source

`provider_options`

provider_options: ProviderOptions

source

`reasoning`

reasoning: ReasoningParam | None

source

`reasoning_reference`

reasoning_reference: ReasoningReferenceParam | None

source

`response_format`

response_format: ResponseFormatParam | None

source

`seed`

seed: int | None

source

`service_tier`

service_tier: ServiceTierEnum | None

source

`stop`

stop: str | list[str] | None

source

`store`

store: bool | None

source

`stream`

stream: bool | None

source

`stream_options`

stream_options: StreamOptionsParam | None

source

`temperature`

temperature: float | None

source

`tool_choice`

tool_choice: ToolChoice | None

source

`tools`

tools: list[FunctionToolParam] | None

source

`top_logprobs`

top_logprobs: int | None

source

`top_p`

top_p: float | None

source

`truncation`

truncation: TruncationEnum | None

source

`user`

user: str | None

source

`validate_single_message_only()`

validate_single_message_only()

source

Validate that only a single message is provided when using list input.

The current implementation only supports single-turn requests with one message. Multi-turn conversations with multiple messages are not yet supported for pixel generation pipelines.

Raises:: ValueError – If input is a list with more than one message.
Returns:: The validated OpenResponsesRequestBody instance.
Return type:: OpenResponsesRequestBody

`validate_streaming_not_supported()`

validate_streaming_not_supported()

source

Validate that streaming is not requested.

Streaming is not currently supported in the MAX implementation of the OpenResponses API. This validator ensures that requests with stream=True are rejected during validation.

Raises:: ValueError – If stream is set to True.
Returns:: The validated OpenResponsesRequestBody instance.
Return type:: OpenResponsesRequestBody

`verbosity`

verbosity: VerbosityEnum | None

source

OpenResponsesRequestBody​

citations​

frequency_penalty​

include​

input​

logit_bias​

logprobs​

max_output_tokens​

metadata​

modalities​

model​

model_config​

n​

parallel_tool_calls​

presence_penalty​

provider_options​

reasoning​

reasoning_reference​

response_format​

seed​

service_tier​

stop​

store​

stream​

stream_options​

temperature​

tool_choice​

tools​

top_logprobs​

top_p​

truncation​

user​

validate_single_message_only()​

validate_streaming_not_supported()​

verbosity​