Bleep API

Authentication

All API requests require authentication using a Bearer token. Include your API key in the Authorization header.

Authorization: Bearer YOUR_API_TOKEN

Getting an API Token

1. Request access via POST /token with your credentials
2. Use the returned token for all subsequent requests

curl -X POST https://api.bleep.stream/token \
  -d "username=YOUR_USERNAME&password=YOUR_PASSWORD"

Response Format

All moderation endpoints return a consistent JSON structure for easy integration.

{
  "id": "modr-abc123",
  "model": "gpt-oss:120b",
  "results": [{
    "flagged": true,
    "categories": {
      "sexual": false,
      "hate": true,
      "violence": false
    },
    "category_scores": {
      "sexual": 0.02,
      "hate": 0.87,
      "violence": 0.12
    }
  }]
}

Categories

Moderation endpoints detect the following content categories:

PII Categories

Additionally, PII detection categories include:

Our most advanced moderation model. GPT-OSS 120B provides superior context understanding and nuanced content analysis with optional reasoning transparency.

curl -X POST https://api.bleep.stream/api/predict/gpt_oss_120b_moderation \
  -H "Authorization: Bearer {token}" \
  -H "Content-Type: multipart/form-data" \
  -F "input_text=Content to moderate" \
  -F "enable_thinking=true"

Fast and efficient moderation using optimized open source LLMs. Great balance of speed and accuracy for text and image content.

curl -X POST https://api.bleep.stream/api/predict/ollama_turbo_moderation \
  -H "Authorization: Bearer {token}" \
  -H "Content-Type: application/json" \
  -d '{"input_text": "Content to analyze"}'

Multi-modal moderation using OpenAI CLIP. Best for high-throughput screening of images and text with low latency requirements.

curl -X POST https://api.bleep.stream/api/predict/clip_moderation \
  -H "Authorization: Bearer {token}" \
  -H "Content-Type: application/json" \
  -d '{"input_text": "Quick content check"}'

Specialized DistilBERT model for detecting offensive language, toxicity, hate speech, and threats.

{
  "text": "your input text",
  "label_scores": {
    "toxic": 0.92,
    "profane": 0.85,
    "insult": 0.12,
    "hate": 0.05,
    ...
  }
}

curl -X POST "https://api.bleep.stream/api/predict/profanity_detection?input_text=test%20message" \
  -H "Authorization: Bearer {token}"

Detect personally identifiable information (PII) using a specialized DistilBERT model trained on PII patterns.

{
  "text": "your input text",
  "label_scores": {
    "EMAIL": 0.95,
    "PERSON": 0.82,
    "TEL": 0.03,
    ...
  }
}

curl -X POST "https://api.bleep.stream/api/predict/personal_detection?input_text=Contact%20me%20at%20test@example.com" \
  -H "Authorization: Bearer {token}"

Analyze images for nudity and explicit content using ONNX-optimized detection models.

curl -X POST https://api.bleep.stream/api/predict/nsfw_detection \
  -H "Authorization: Bearer {token}" \
  -F "image=@/path/to/image.jpg"

Get detailed classification of NSFW content type for more granular filtering decisions.

curl -X POST https://api.bleep.stream/api/predict/nsfw_type \
  -H "Authorization: Bearer {token}" \
  -F "image=@/path/to/image.jpg"