Fix: Moderations endpoint now respects api_base configuration parameter

[Sameerlite]

Title

Fix: Moderations endpoint now respects api_base configuration parameter

Relevant issues

Fixes LIT-1385

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/litellm/ directory, Adding at least 1 test is a hard requirement - see details

• I have added a screenshot of my new test passing locally

• My PR passes all unit tests on make test-unit

• My PR's scope is as isolated as possible, it only solves 1 specific problem

Type

🐛 Bug Fix

Changes

Problem

The Moderations endpoint (/v1/moderations) was not respecting the api_base configuration parameter. When making requests with a custom api_base in the config (e.g., for OpenAI Data Residency using https://us.api.openai.com), requests were still being sent to the default OpenAI endpoint.

Root Cause

Both the synchronous moderation() and asynchronous amoderation() functions in litellm/main.py were not passing the api_base parameter when creating the OpenAI client.

Solution

1. Fixed amoderation() (async function) - Line 5335-5366

• Moved optional_params extraction and get_llm_provider() call before creating the OpenAI client
• Added api_base=optional_params.api_base or _dynamic_api_base parameter when calling _get_openai_client()

2. Fixed moderation() (sync function) - Line 5290-5319

• Extracted api_base from kwargs
• Conditionally added base_url parameter to OpenAI client initialization when api_base is provided

3. Added regression test - test_moderation_endpoint_with_api_base()

• Location: tests/router_unit_tests/test_router_endpoints.py
• Verifies that custom api_base is correctly passed to the OpenAI client
• Uses mocking to ensure the parameter flows through correctly

- model_name: "*"
    litellm_params:
      model: "openai/*"
      api_base: http://host.docker.internal:8080/
      api_key: dummy
  - model_name: "openai/*"
    litellm_params:
      model: "openai/*"
      api_base: http://host.docker.internal:8080/
      api_key: dummy

curl http://localhost:4000/v1/moderations \  
 -X POST \
 -H "Content-Type: application/json" \
 -H "Authorization: Bearer $API_KEY" \
 -d '{
  "model": "openai/omni-moderation-latest",
  "input": "...text to classify goes here..."
 }'

It shows it is trying to use http://host.docker.internal:8080/

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Preview	Comments	Updated (UTC)
litellm	Ready	Preview	Comment	Oct 30, 2025 7:53am

[ishaan-jaff]

LGTM