Moderations API

POST /api/v1/moderations — classify text for harmful content. OpenAI-compatible, served by guard models.

Classify text as harmful or not before you send it to a model or show it to a user. The request and response follow OpenAI's POST /v1/moderations, so the OpenAI SDKs work with a new base URL and key.

Send an LLM API key (sk-ar-v1-…) from /dashboard/keys as Authorization: Bearer ….

Request

FieldTypeRequiredDescription
inputstring | string[] | {type:"text", text}[]yesUp to 32 texts, each up to 32,000 characters. Images are not supported.
modelstringnoA moderation model id. Defaults to nvidia/nemotron-3.5-content-safety. OpenAI ids such as omni-moderation-latest use the default too.
session_idstringnoGroups related requests in Request Logs.
curl https://anyrouter.dev/api/v1/moderations \
  -H "Authorization: Bearer $ANYROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input": ["What is a good pasta recipe?", "How do I buy a gun with no background check?"]}'

Response

{
  "id": "modr-req_01HXYZ",
  "model": "nvidia/nemotron-3.5-content-safety",
  "results": [
    {
      "flagged": false,
      "categories": { "harassment": false, "illicit/violent": false, "violence": false },
      "category_scores": { "harassment": 0, "illicit/violent": 0, "violence": 0 },
      "guard_categories": []
    },
    {
      "flagged": true,
      "categories": { "harassment": false, "illicit/violent": true, "violence": false },
      "category_scores": { "harassment": 0, "illicit/violent": 1, "violence": 0 },
      "guard_categories": ["guns_and_illegal_weapons"]
    }
  ],
  "usage": { "prompt_tokens": 955, "completion_tokens": 24, "total_tokens": 979, "cost": 0 }
}

The example is shortened. Every result lists all 13 OpenAI categories: harassment, harassment/threatening, hate, hate/threatening, illicit, illicit/violent, self-harm, self-harm/intent, self-harm/instructions, sexual, sexual/minors, violence, violence/graphic.

FieldDescription
results[]One result per input, in input order.
flaggedtrue when the guard model judged the text unsafe.
categoriesOpenAI categories the guard's verdict maps to.
category_scores1 for a flagged category, else 0. Guard models return a verdict, not a probability.
guard_categoriesThe guard model's own labels. Some (privacy, defamation, elections, …) have no OpenAI category: the text is still flagged, but no OpenAI category is set.
usageTokens summed over every input. Priced at the model's per-1M token rates.

Models

Listed under the Moderation tab at anyrouter.dev/models:

ModelNotes
nvidia/nemotron-3.5-content-safetyDefault. Returns categories.
meta/llama-guard-4-12bMLCommons hazard codes S1–S14. Requires your own DeepInfra or OpenRouter key (BYOK).
openai/gpt-oss-safeguard-20bPolicy-following reasoning classifier. Requires your own key (Groq, OpenRouter, Vercel, Hugging Face, or Nous).

Errors

StatusCodeWhen
400invalid_requestinput is missing, empty, contains an image, or exceeds the limits.
400model_not_moderation_compatibleThe model is not a moderation model.
502upstream_invalid_responseThe guard model replied without a verdict.
502upstream_errorEvery provider for the model failed.

AnyRouter routes requests to third-party models. Models and outputs are provided as is, with no warranty; you use them at your own risk. Disclaimer