
Automatically rate images as G, PG-13, R, or Adult based on nudity, violence, and language. Ensure age-appropriate content across your platform.
The Bynn Content Rating model analyzes images to assign age-appropriate ratings based on nudity, violence, and language. This model classifies content as General Audience (PG), Teen 13+ (PG-13), Teen 16+ (16+), or Adult (R).
Parents trust platforms to protect their children from age-inappropriate content. Regulators mandate age-gating for mature material. Yet the volume of user-generated content makes manual rating impossible, while binary "safe/unsafe" classifications fail to capture the nuanced spectrum of content appropriateness.
A PG-13 action movie differs from R-rated horror. Mild profanity is acceptable for teens but not children. Swimwear at a beach is family-friendly; the same exposure in a sexualized context is not. Platforms need granular, consistent rating that mirrors established systems like MPAA ratings—providing parents and users the information they need to make informed viewing decisions.
When provided with an image, the model evaluates multiple content dimensions including nudity levels, violence, and visible text with profanity. It applies a hierarchical classification system that escalates ratings based on the most restrictive content present.
Achieving 93.0% accuracy, the model uses Bynn's Visual Language Model technology to perform contextual visual reasoning with sophisticated understanding of content appropriateness standards.
The model applies five core rules for consistent classification:
The API returns a structured JSON response containing:
Family-safe content: wholesome, educational, everyday content. People fully covered or in normal attire with no sexual emphasis. No wounds, blood, fighting, or crude text.
Safe content that should NOT be escalated:
Content suitable for teens 13 and older:
Content suitable for older teens 16 and above:
Adult-only content:
| Metric | Value |
|---|---|
| Classification Accuracy | 93.0% |
| Average Response Time | 15,000ms |
| Max File Size | 20MB |
| Supported Formats | GIF, JPEG, JPG, PNG, WebP |
Important Considerations:
This model provides probability-based classifications, not official content ratings.
Best Practice: Combine rating results with human review and consider the confidence score when making content distribution decisions.
Vision Language Model for image/video understanding with reasoning
media_typestringType of media being sent: 'image' or 'video'. Auto-detected if not specified.
imageimage_urlstringURL of image to analyze
https://example.com/image.jpgbase64_imagestringBase64-encoded image data
video_urlstringURL of video to analyze
https://example.com/video.mp4base64_videostringBase64-encoded video data
Structured Content Rating response
responseobjectStructured response from the model
ratingstringgeneral_audienceteen_13_plusteen_16_plusadult_18_plusreasonstringconfidencestringlowmediumhighthinkingstringChain-of-thought reasoning from the model (may be empty)
{
"model": "content-rating",
"image_url": "https://example.com/image.jpg"
}{
"inference_id": "inf_abc123def456",
"model_id": "content_rating",
"model_name": "Content Rating",
"moderation_type": "image",
"status": "completed",
"result": {
"response": {
"rating": "general_audience",
"reason": "example_reason",
"confidence": "low"
},
"thinking": ""
}
}429 HTTP error code along with an error message. You should then retry with an exponential back-off strategy, meaning that you should retry after 4 seconds, then 8 seconds, then 16 seconds, etc.Integrate Content Rating into your application today with our easy-to-use API.