Skip to main content
GET
Get Feature Info

Path Parameters

feature
string
required
subfeature
string
required

Query Parameters

format
enum<string>
default:simplified

Schema format: 'simplified' for flat readable format, 'json_schema' for full Pydantic JSON Schema

Available options:
simplified,
json_schema
zero_data_retention
boolean | null

true keeps only models whose provider keeps no prompts or outputs after the response (Zero Data Retention); false keeps only those known to keep them. Models not verified yet (zero_data_retention: null) match neither.

Response

Successful Response

Detailed response for a specific subfeature.

feature
string
required
feature_fullname
string
required
subfeature
string
required
subfeature_fullname
string
required
mode
enum<string>
required
Available options:
sync,
async
endpoints
Endpoints · object
required
request_schema
Request Schema · object
required

Schema of the request body sent to endpoints.create. The fields of input_schema go inside its input field, and model takes one of the models ids.

input_schema
Input Schema · object
required
output_schema
Output Schema · object
required
models
Models · object[]
required

Models available for this subfeature: the model id to send, its pricing and regions, and documentation_url, the provider's documentation for that model (the model's own page when there is one, else the feature's, else the provider's; null when none is known). requires_reference_images is true for an edit-only model, which fails on a request without reference_images, and requires_video for a model that edits a source video, which fails without video.

description
string | null
provider_params
Provider Params · object | null

The provider_params each provider accepts, for subfeatures that restrict them; any other parameter is rejected with 400. Each entry is a parameter path in the provider's own request format, nested keys joined with dots (e.g. videogenerationconfig.seed), matched case-insensitively: send the keys in the provider's own casing ({"videoGenerationConfig": {"seed": 7}}). An empty list means the provider accepts none; null on subfeatures that publish no list.