An IP stresser sends synthetic traffic at one host for a fixed window, then reports what happened: requests per second, p50/p95/p99 latency, errors and the rate where the curve stops being linear. It measures capacity, not security. Nothing here finds an injection flaw or a weak password.
The term is also used by panels that will aim at any address a customer types in. That is a different product with the same traffic shape, and the difference is authorisation: a DNS TXT record you can publish, checked again at run start. If a provider does not check, it is not selling you a test.
Start free, read the FAQ, or pick a stresser page from the grid below.