SnozeDevelopers
API referenceModel sources

List the models a model source exposes

Decrypts the stored key after re-checking access and asks the provider for its current model catalog, returning each model's id, context length, and provider-reported pricing. Requires `models:read`. Returns a not-found error for an unknown source and a fetch error when the provider is unreachable or rejects the key.

GET
/v1/workspaces/{workspaceId}/model-sources/{modelSourceId}/models

Authorization

apiKey
headerAuthorizationBearer <snoze_sk_…>

Workspace API key created in Workspace settings → API keys. Scopes on the key bound what it can read and write.

Path Parameters

workspaceId*string
modelSourceId*string

Response Body

Success

application/json
  1. response
data*
Example Request
curl -X GET 'https://api.snoze.dev/v1/workspaces/string/model-sources/string/models'
Example Response
{  "data": {    "models": [      {        "id": "string",        "label": "string",        "description": "string",        "contextLength": 0,        "pricing": {          "prompt": "string",          "completion": "string",          "inputCacheRead": "string",          "inputCacheWrite": "string",          "request": "string",          "tiers": [            {              "minPromptTokens": 0,              "prompt": "string",              "completion": "string",              "inputCacheRead": "string",              "inputCacheWrite": "string"            }          ]        },        "metadata": {          "inputModalities": [            "string"          ],          "outputModalities": [            "string"          ],          "supportedParameters": [            "string"          ],          "reasoning": {            "mandatory": true,            "supportedEfforts": [              "string"            ],            "defaultEffort": "string",            "defaultEnabled": true          },          "contextLength": 0,          "maxOutputTokens": 0,          "pricing": {            "prompt": "string",            "completion": "string",            "inputCacheRead": "string",            "inputCacheWrite": "string",            "request": "string",            "tiers": [              {                "minPromptTokens": 0,                "prompt": "string",                "completion": "string",                "inputCacheRead": "string",                "inputCacheWrite": "string"              }            ]          },          "created": 0,          "expirationDate": "string",          "canonicalSlug": "string"        }      }    ]  }}

Delete a model source

Removes the provider key and its model selection from the workspace, so features relying on it fall back to the platform's own models. Requires `models:write` and returns a not-found error for an unknown source.

Unfurl an external URL into a link preview

Fetches the page named by the `url` query parameter server-side and returns its title, description, preview image, site name, and icon, parsed from Open Graph, Twitter card, and plain HTML tags. Requires the `workspace:read` permission. Only public http(s) URLs are resolved: other schemes, `localhost`, and private or loopback IP addresses are refused as validation errors, redirects are followed at most three times with each hop re-checked, the response body is read up to 512 KB, and non-HTML responses come back as a bare URL. Results are cached for a day per URL, and a missing `url`, an unreachable host, or a non-success response is reported as a validation or fetch error.