Yes you can load it with no need of passing through GCS. Using this tutorial as a guide (which uses JSON files) brings us to this Source Code (used by the Cloud Function in the tutorial), as you can see all it does is load the bucket file into a blob and then to a json :
row = json.loads(blob.download_as_string())
After that is just a matter of using the insert_rows_json method :
BQ.insert_rows_json(table, json_rows=[row],
row_ids=[file_name],
retry=retry.Retry(deadline=30))
So, you can just send from your API the requests with a POST to a HTTP Cloud Function and that function use the above code to load the json fo BigQuery.