Safety Checker
All predictions run with the safety checker enabled by default (enable_safety_checker: true). When enabled, generated content is filtered for NSFW material. If the output gets flagged, the prediction returns a filtered result or an error.
To disable the safety checker, pass enable_safety_checker: false in the input object of your prediction request.
Usage
curl -X POST https://api.eachlabs.ai/v1/prediction \
-H "Content-Type: application/json" \
-H "X-API-Key: YOUR_API_KEY" \
-d '{
"model": "wan-v2-6-text-to-video",
"version": "1.0",
"input": {
"prompt": "your prompt here",
"enable_safety_checker": false
}
}'
Supported Models
The enable_safety_checker parameter is not available on all models. Only certain models let you toggle the safety filter. Models that don’t support it will just quietly ignore the parameter.
| Model | Slug | Type |
|---|
| Wan v2.6 Text-to-Video | wan-v2-6-text-to-video | Video Generation |
| Wan v2.6 Image-to-Video | wan-v2-6-image-to-video | Video Generation |
| Seedream v4.5 | seedream-v4-5-text-to-image | Image Generation |
This list might not cover everything! Model support for this parameter can change as we add or update models. Check the model’s input schema via GET /v1/model?slug={slug} to confirm whether enable_safety_checker is accepted.
Behavior Summary
enable_safety_checker | Supported Model | Unsupported Model |
|---|
true (default) | NSFW content filtered | NSFW content filtered |
false | NSFW filter disabled | Parameter ignored, filter stays on |
| Not provided | Same as true | Same as true |
each::sense
When using each::sense, pass enable_safety_checker as a top-level request field instead of inside input:
{
"messages": [{"role": "user", "content": "your prompt here"}],
"enable_safety_checker": false
}
The safety checker is a model-level feature. each::labs does not add an additional content filter on top of the model’s own safety system.
