Seems like there's room for some common helper functions, such as checking that error != nil (or failing the test) and checking that the response is a 2xx (or failing the test).
Bulk operations test looks to be missing bulk-create and bulk-partial-update (PATCH) operations coverage
Single operations test looks to be missing partial-update (PATCH) coverage
Probably don't need to test CRUD for all models but it's worth thinking about testing the various bespoke endpoints (devices/napalm, jobs/run, prefixes/available-prefixes, etc.) to confirm that they appear to work correctly as generated.