by how much? I don't understand the cost model here at all.
AIUI the idea is to ratelimit each "solution". A normal human's browser only needs to "solve" once. A LLM crawler either needs to slow down (= objective achieved) or solve the puzzle n times to get n × the request rate.
AIUI the idea is to ratelimit each "solution". A normal human's browser only needs to "solve" once. A LLM crawler either needs to slow down (= objective achieved) or solve the puzzle n times to get n × the request rate.