The Synas models
Atugil Intelligence publishes two open-weight vision models. One generates images. The other checks them.
| Synas Image 001 | Synas Detect 001 | |
|---|---|---|
| Job | Generates images from text and edits existing images | Estimates whether an image was AI-generated or edited |
| You give it | A text prompt, and optionally an image to edit | An image |
| You get back | An image | A calibrated score and its confidence |
| Built around | Deterministic seeds and style locks | Diffusion fingerprints, inpainting seams and metadata mismatches |
| Weights | Open-weight | Open-weight |
| Best for | Creative and research pipelines that need repeatable images | Review queues, trust and safety, research, and anyone checking an image before using it |
| Access | Waitlist | Waitlist |
Which one do you need?
You make images
Start with Image 001. Seeds and style locks give you repeatable results you can log, review and regenerate.
You receive or publish images
Start with Detect 001. It gives you a calibrated score and a confidence to feed into your review process.
You do both, or you research this
Use them together. Generate edits with Image 001 and see whether Detect 001 flags them. Testing your checks against your own generator is the quickest way to learn where they fail.
You want to understand the ideas first
Read the guide to how AI image detection works, or the difference between open-weight and open-source.
Get early access
Join the waitlist and we'll email you when the models and their licenses are released. One email, no spam, remove yourself any time.