Implementation:Openai Openai python Moderations Resource

Knowledge Sources	Openai_Openai_python
Domains	Content_Safety
Last Updated	2026-02-15 00:00 GMT

Overview

Concrete tool for the Moderation API resource provided by the openai-python SDK.

Description

The moderations module implements the Moderations (sync) and AsyncModerations (async) resource classes for the POST /moderations endpoint. The create() method accepts an input parameter that can be a single string, a sequence of strings, or an iterable of ModerationMultiModalInputParam objects (for combined text and image moderation). An optional model parameter selects the moderation model. The method transforms the input using maybe_transform() against ModerationCreateParams and returns a ModerationCreateResponse containing flagged categories and scores. Both classes provide with_raw_response and with_streaming_response cached properties that wrap the resource in ModerationsWithRawResponse, AsyncModerationsWithRawResponse, ModerationsWithStreamingResponse, or AsyncModerationsWithStreamingResponse for accessing raw HTTP response data or streaming responses. This file is auto-generated from the OpenAPI spec by Stainless.

Usage

Use this resource to classify text and/or image content for potentially harmful categories (hate, violence, self-harm, sexual, etc.) before displaying or processing user-generated content.

Code Reference

Source Location

• Repository: openai-python
• File: src/openai/resources/moderations.py
• Lines: 1-197

Signature

class Moderations(SyncAPIResource):
    @cached_property
    def with_raw_response(self) -> ModerationsWithRawResponse: ...
    @cached_property
    def with_streaming_response(self) -> ModerationsWithStreamingResponse: ...

    def create(
        self,
        *,
        input: Union[str, SequenceNotStr[str], Iterable[ModerationMultiModalInputParam]],
        model: Union[str, ModerationModel] | Omit = omit,
        extra_headers: Headers | None = None,
        extra_query: Query | None = None,
        extra_body: Body | None = None,
        timeout: float | httpx.Timeout | None | NotGiven = not_given,
    ) -> ModerationCreateResponse: ...

class AsyncModerations(AsyncAPIResource):
    async def create(
        self,
        *,
        input: Union[str, SequenceNotStr[str], Iterable[ModerationMultiModalInputParam]],
        model: Union[str, ModerationModel] | Omit = omit,
        ...
    ) -> ModerationCreateResponse: ...

Import

from openai.resources.moderations import Moderations, AsyncModerations

I/O Contract

Inputs

Name	Type	Required	Description
input	str / SequenceNotStr[str] / Iterable[ModerationMultiModalInputParam]	Yes	Text or multi-modal content to classify
model	str / ModerationModel / Omit	No	Moderation model to use (omit for default)
extra_headers	Headers or None	No	Additional HTTP headers
extra_query	Query or None	No	Additional query parameters
extra_body	Body or None	No	Additional JSON body properties
timeout	float / httpx.Timeout / None / NotGiven	No	Request timeout override

Outputs

Name	Type	Description
create() result	ModerationCreateResponse	Contains moderation results with flagged categories and confidence scores

Usage Examples

Basic Usage

from openai import OpenAI

client = OpenAI()

# Moderate a single string
response = client.moderations.create(input="Some text to check")
print(response.results[0].flagged)       # True/False
print(response.results[0].categories)    # Category flags
print(response.results[0].category_scores)  # Confidence scores

# Moderate multiple strings
response = client.moderations.create(
    input=["First text", "Second text"],
    model="omni-moderation-latest",
)
for result in response.results:
    print(result.flagged)

# Access raw response headers
raw = client.moderations.with_raw_response.create(input="test")
print(raw.headers["x-request-id"])
moderation = raw.parse()

Related Pages

• Environment:Openai_Openai_python_Python_3_9_Plus