Read an inference option

GET

/inferences/{id}

const url = 'https://api.canopyhq.dev/inferences/example';
const options = {method: 'GET', headers: {Authorization: 'Bearer <token>'}};

try {
  const response = await fetch(url, options);
  const data = await response.json();
  console.log(data);
} catch (error) {
  console.error(error);
}

curl --request GET \
  --url https://api.canopyhq.dev/inferences/example \
  --header 'Authorization: Bearer <token>'

Authorizations

console_token

Parameters

Path Parameters

required

string

The option, by the name the tenant gave it.

Responses

200

The option.

application/json

A preset on a provider: the model, the generation defaults a session’s dial overrides, when to compact, and how often to retry. Several presets may run on one provider; consent selects a set of them.

object

compact_at

Compact before the next request after a completion reaches this prompt-token count.

integer format: int64

model

required

The provider’s model slug, such as anthropic/claude-opus-5.

string

name

A readable name, such as fast or smart.

string

params

Defaults the dial’s own parameters override field by field.

object

effort

One of:

null
string

null

max_tokens

integer | null format: int64

stop

Array<string>

temperature

number | null format: double

top_p

number | null format: double

provider

required

One of:

string
object

A provider’s name within its tenant.

string

retry

How often a transient provider failure is attempted, with doubling waits.

object

attempts

required

integer format: int32

backoff

required

string

Example

{
  "params": {
    "effort": "low"
  },
  "provider": {
    "kind": "simulated"
  }
}

401

No console token, or one that did not verify.

404

The tenant owns no option by that name.