Models & Safety
See the image models currently available in Magggic, their exact provider model IDs, and the safety checks applied before generation and publication.
Last updated: August 10, 2026Currently available models
Magggic currently provides image generation only. Video generation is not offered. Each product name below maps directly to the model ID used by the generation service.
Seedream 5.0 Pro
ByteDance Seedream
- Provider model ID
doubao-seedream-5-0-pro- Access path
- Accessed through APIMart
- Input
- Text prompt with up to 4 optional reference images
- Output
- Static image
GPT Image 2.0
OpenAI
- Provider model ID
gpt-image-2-official- Access path
- Accessed through APIMart
- Input
- Text prompt with up to 4 optional reference images
- Output
- Static image
Nano Banana Pro
Google Gemini
- Provider model ID
gemini-3-pro-image-preview- Access path
- Accessed through APIMart
- Input
- Text prompt with up to 4 optional reference images
- Output
- Static image
Content safety flow
Safety checks are applied at three real stages: request intake, model generation, and public publication.
- 1
Prompt scan before generation
Before a task is created or credits are charged, the prompt is sent to Waffo Prompt Sift. Only an allow result proceeds; review, block, or an unavailable scan service stops the request.
- 2
Model-provider safety controls
Allowed requests remain subject to APIMart and the selected model provider's safety rules. GPT Image 2.0 uses moderation: auto. A provider may refuse a high-risk request or return no output.
- 3
Human review before publication
Generated images remain private by default. When a user requests publication, the artwork enters a human review queue and does not appear in Explore until approved.
Policies and reporting
To report suspected prohibited content, send the artwork link, reason, and supporting context to support@magggic.com.