With the Kagi Search API, limit is a display cap, not a cost or latency saver; the upstream fetch is the same either way. Set it to control how much you read, not how much you spend.

Context: From Kagi's official OpenAPI docs for the search request. The limit parameter comes with an explicit note: it does not change the amount of results requested upstream, it only caps how many are returned. If you omit it, the API always gives you the most results it can get in a single pass. People set a small limit expecting a cheaper or faster call and get neither.