You could use the subjects endpoint with a level filter.
GET https://api.wanikani.com/v2/subjects?levels=<level>
You’ll need to count them yourself though, looking at the object property for each item.
The possible values are radical, kanji, and vocabulary
That’ll do it with one request. If you want to be lazy, and spend 3 requests, you can just filter the subjects by both level and type, and use the response.total_count value. For example:
GET https://api.wanikani.com/v2/subjects?levels=1&types=kanji
{
"object": "collection",
"url": "https://api.wanikani.com/v2/subjects?levels=1&types=kanji",
"pages": {
"per_page": 1000,
"next_url": null,
"previous_url": null
},
"total_count": 18,
"data_updated_at": "2020-06-17T21:10:46.748116Z",
"data": [
...
]
}