Implementation:Openai Openai python Response Create Params

Knowledge Sources	Openai_Openai_python
Domains	API_Types, Python
Last Updated	2026-02-15 00:00 GMT

Overview

Concrete type definitions for Responses API creation parameters provided by the openai-python SDK.

Description

The response_create_params.py module defines the TypedDict parameter types used when creating a new response via the OpenAI Responses API. It contains three primary types organized in an inheritance hierarchy:

ResponseCreateParamsBase -- The base TypedDict that defines all shared parameters for response creation, including model, input, instructions, tools, tool_choice, temperature, top_p, max_output_tokens, and many more.
ResponseCreateParamsNonStreaming -- Extends the base with stream: Optional[Literal[False]] for non-streaming requests.
ResponseCreateParamsStreaming -- Extends the base with stream: Required[Literal[True]] for streaming requests.

The final ResponseCreateParams is a Union of the streaming and non-streaming variants.

Additional helper types defined in this module include:

ContextManagement -- TypedDict for context management (compaction) configuration.
Conversation -- TypeAlias for conversation references (Union[str, ResponseConversationParam]).
StreamOptions -- TypedDict for streaming configuration options (e.g., obfuscation).
ToolChoice -- TypeAlias union of all tool choice parameter variants.

This file is auto-generated from the OpenAI API spec by the Stainless code generation toolchain.

Usage

Use these types when constructing parameters for client.responses.create() calls, or when annotating function signatures that accept response creation configuration. The streaming vs. non-streaming distinction determines whether the API returns a single response or a stream of server-sent events.

Code Reference

Source Location

Repository: openai-python
File: src/openai/types/responses/response_create_params.py
Lines: 1-348

Signature

class ResponseCreateParamsBase(TypedDict, total=False):
    background: Optional[bool]
    context_management: Optional[Iterable[ContextManagement]]
    conversation: Optional[Conversation]
    include: Optional[List[ResponseIncludable]]
    input: Union[str, ResponseInputParam]
    instructions: Optional[str]
    max_output_tokens: Optional[int]
    max_tool_calls: Optional[int]
    metadata: Optional[Metadata]
    model: ResponsesModel
    parallel_tool_calls: Optional[bool]
    previous_response_id: Optional[str]
    prompt: Optional[ResponsePromptParam]
    prompt_cache_key: str
    prompt_cache_retention: Optional[Literal["in-memory", "24h"]]
    reasoning: Optional[Reasoning]
    safety_identifier: str
    service_tier: Optional[Literal["auto", "default", "flex", "scale", "priority"]]
    store: Optional[bool]
    stream_options: Optional[StreamOptions]
    temperature: Optional[float]
    text: ResponseTextConfigParam
    tool_choice: ToolChoice
    tools: Iterable[ToolParam]
    top_logprobs: Optional[int]
    top_p: Optional[float]
    truncation: Optional[Literal["auto", "disabled"]]
    user: str

class ResponseCreateParamsNonStreaming(ResponseCreateParamsBase, total=False):
    stream: Optional[Literal[False]]

class ResponseCreateParamsStreaming(ResponseCreateParamsBase):
    stream: Required[Literal[True]]

ResponseCreateParams = Union[ResponseCreateParamsNonStreaming, ResponseCreateParamsStreaming]

Import

from openai.types.responses.response_create_params import (
    ResponseCreateParamsBase,
    ResponseCreateParamsNonStreaming,
    ResponseCreateParamsStreaming,
)
# Or via the top-level re-export:
from openai.types.responses import ResponseCreateParams

I/O Contract

Inputs

Name	Type	Required	Description
model	ResponsesModel	No (total=False)	Model ID to use, e.g. `"gpt-4o"` or `"o3"`
input	Union[str, ResponseInputParam]	No	Text, image, or file inputs to generate a response from
instructions	Optional[str]	No	System/developer message inserted into the model context
tools	Iterable[ToolParam]	No	Array of tools the model may call (functions, web search, MCP, etc.)
tool_choice	ToolChoice	No	How the model should select tools
temperature	Optional[float]	No	Sampling temperature between 0 and 2
top_p	Optional[float]	No	Nucleus sampling probability mass
max_output_tokens	Optional[int]	No	Upper bound for generated tokens including reasoning tokens
max_tool_calls	Optional[int]	No	Maximum total built-in tool calls allowed
stream	Literal[True] / Optional[Literal[False]]	Varies	Controls streaming vs. non-streaming mode
store	Optional[bool]	No	Whether to store the response for later retrieval
previous_response_id	Optional[str]	No	ID of previous response for multi-turn conversations
conversation	Optional[Conversation]	No	Conversation reference for stateful interactions
truncation	Optional[Literal["auto", "disabled"]]	No	Truncation strategy for context overflow
reasoning	Optional[Reasoning]	No	Configuration for reasoning models (o-series, gpt-5)
service_tier	Optional[Literal["auto", "default", "flex", "scale", "priority"]]	No	Processing tier for the request
background	Optional[bool]	No	Whether to run the response in background mode
metadata	Optional[Metadata]	No	Up to 16 key-value pairs for object metadata
include	Optional[List[ResponseIncludable]]	No	Additional output data to include (sources, logprobs, etc.)
text	ResponseTextConfigParam	No	Configuration for text response format
prompt	Optional[ResponsePromptParam]	No	Reference to a reusable prompt template
prompt_cache_key	str	No	Cache key for prompt caching optimization
prompt_cache_retention	Optional[Literal["in-memory", "24h"]]	No	Retention policy for prompt cache
safety_identifier	str	No	Stable identifier for abuse detection
parallel_tool_calls	Optional[bool]	No	Whether to allow parallel tool execution
top_logprobs	Optional[int]	No	Number of most likely tokens to return (0-20)
user	str	No	Deprecated; replaced by safety_identifier and prompt_cache_key

Outputs

Name	Type	Description
ResponseCreateParamsBase	TypedDict	Base parameter dictionary with all shared fields
ResponseCreateParamsNonStreaming	TypedDict	Non-streaming variant (stream=False or omitted)
ResponseCreateParamsStreaming	TypedDict	Streaming variant (stream=True required)
ResponseCreateParams	Union	Union of streaming and non-streaming params
ToolChoice	TypeAlias	Union of ToolChoiceOptions, ToolChoiceAllowedParam, ToolChoiceFunctionParam, ToolChoiceMcpParam, ToolChoiceCustomParam, ToolChoiceApplyPatchParam, ToolChoiceShellParam, ToolChoiceTypesParam
Conversation	TypeAlias	Union[str, ResponseConversationParam]
StreamOptions	TypedDict	Streaming options with include_obfuscation flag
ContextManagement	TypedDict	Context management entry with type and compact_threshold

Usage Examples

Basic Non-Streaming Request

from openai import OpenAI
from openai.types.responses import ResponseCreateParams

client = OpenAI()

response = client.responses.create(
    model="gpt-4o",
    input="Explain quantum computing in simple terms.",
    instructions="You are a helpful science teacher.",
    temperature=0.7,
    max_output_tokens=500,
)
print(response.output)

Streaming Request

from openai import OpenAI

client = OpenAI()

stream = client.responses.create(
    model="gpt-4o",
    input="Write a short poem about Python programming.",
    stream=True,
)
for event in stream:
    print(event)

With Tools and Tool Choice

from openai import OpenAI

client = OpenAI()

response = client.responses.create(
    model="gpt-4o",
    input="What is the weather in San Francisco?",
    tools=[
        {
            "type": "function",
            "name": "get_weather",
            "description": "Get the current weather for a location.",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {"type": "string"},
                },
                "required": ["location"],
            },
        }
    ],
    tool_choice="auto",
)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment