Classify unsafe text and images

Moderations

Classify unsafe text and image inputs with the OpenAI-compatible /v1/moderations API.

Get startedRead the docs
POST /v1/moderations
I want to harm someone.
  1. 1

    violence

    flagged
  2. 2

    harassment

    0.34
  3. 3

    self-harm

    0.03
01

Text & images

Multimodal safety checks.

02

Category scores

Know why it was flagged.

03

OpenAI format

Existing code just works.

04

Pre-check

Filter before the big model.

Built for

  • UGC platforms

    Screen posts and uploads

  • Chat safety

    Check prompts before answering

  • Policy audits

    Score historical content

Check an input
terminal
1curl https://api.llmgateway.io/v1/moderations \2  -H "Authorization: Bearer $LLM_GATEWAY_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "input": "I want to harm someone."6  }'
Read docs

More in models & modalities

All features

Transcription

Speech to text with word timestamps

Realtime API

Speech-to-speech over WebSockets

Vision

Send images, get answers

Start using Moderations today

One API key. Every model. Live in minutes.

Start buildingBrowse models