Pruna API Dev Portal

1. Log in

1. Go to dashboard.pruna.ai.
2. Log in with the email and temporary password we sent you.
3. The password we provided is temporary — once you’re in, go to your "profile" then "security settings" and change it.

2. Set up your workspace & team (optional but useful)

• In the Team tab:
  • Invite teammates by email.
  • They’ll get an invite and can create their own API keys.
  • As an admin, you can remove members and see pending invitations.

3. Create your API key

1. Go to the API Keys tab.
2. Click “Create API key”, give it a name (e.g. prod-backend, staging, alex-local).
3. Copy it immediately — you won’t see it again after closing the popup.
4. If you lose it:
   • Use “Rotate”: this invalidates the old key and issues a new one (no need to delete + recreate from scratch).
5. Use it in your requests as an HTTP header:
   • Header name: apikey (lowercase, no spaces)
   • Value: your API key
   • You also need this header when downloading generated images/videos from the generation_url.

4. Read the Quickstart in the docs

1. From the dashboard, click “Docs” (top-right).
2. Go to Quickstart:
   • How to call the /predictions endpoint
   • The difference between synchronous (try-sync: true) and asynchronous (polling a status route)
   • How to upload images for edit/video models

5. Choose and use the model you want

• In Docs → Models, you’ll see all available models:
  • P-Image — flagship text-to-image
  • P-Image-edit — flagship image editing
  • Other performance models
• Each model page contains:
  • A minimal code snippet for calling the API
  • The list of parameters (aspect ratio, guidance, frames for video, etc.)
  • The pricing per request

Recommended defaults:

• For fast models (like P-Image):
  → You can use try-sync: true (synchronous) for a simple integration.
• For heavier models (video, big resolutions):
  → Use asynchronous flow: send request, then poll the status endpoint until it returns succeeded, then download from generation_url.