this is related to https://forum.inaturalist.org/t/label-species-that-are-included-in-the-cv-training-model/21831.
right now, i think the way to do what you’re describing would be 2 parts. the first part would be to get a list of species. you can do that various ways. once you have your list, you can get the taxon details from the API via /v1/taxa/{id}, and that response will include a vision field which tells you whether the taxon is included in the computer vision model. you can parse those results based on that field to get just the records where vision=true for those that are included in the model or vision=false for those that are not part of the model.
just for example, here’s the request URL to get taxon information for a couple of species of North American bears: https://api.inaturalist.org/v1/taxa/41641,41638.