Skip to main content

​
How it works

Each key has two optional fields:
  • allowedModels — if non-empty, only these models are usable
  • blockedModels — always rejected; takes precedence over allowed
Empty / null on both = all models allowed (default behavior).

​
Set via dashboard

dolphy.chat/settings/api-keys → Manage on the key → enter comma-separated model ids.

​
Set via API

curl -X PATCH https://dolphy.chat/api/account/api-keys/dpy_live_abc12345 \
  -H "Authorization: Bearer <firebase-id-token>" \
  -H "Content-Type: application/json" \
  -d '{
    "allowedModels": ["lustify-v8", "dolphy-chat-1"],
    "blockedModels": ["nano-banana-pro", "gpt-image-2"]
  }'
Pass null or [] to clear either list.

​
Common patterns

​
”Only the cheap models on the dev key”

{
  "allowedModels": [
    "lustify-v8",
    "lustify-v7",
    "chroma",
    "z-image-turbo",
    "wai-Illustrious",
    "seedream-v4",
    "hidream"
  ]
}
All 1-credit/image models. Caps surprise spend on premium models.

​
”Restrict to a curated set for a customer-facing app”

{
  "allowedModels": ["flux-2-pro", "seedream-v4", "recraft-v4", "gpt-image-2"]
}
Only models in allowedModels will be accepted by this key. Anything else returns 403.

​
”Just chat for the AI agent”

{ "allowedModels": ["dolphy-chat-1"] }
Blocks every image and video model. Only chat allowed.

​
When a request hits a blocked model

{
  "error": {
    "message": "This API key is not permitted to use model \"nano-banana-pro\".",
    "type": "model_forbidden",
    "code": null,
    "param": null
  }
}
Status: 403. The customer should either change the model in their request or update the key’s permissions.

​
Listing model ids

GET /v1/models returns all available model ids with their pricing and capabilities. Use this to populate a model picker in your client.