Captcha worker pay
Captcha worker pay is what a service pays the person clearing a challenge. Proxy Compare reads it from the recruitment pages of the 10 services it tracks, 3 Oct 2026, and publishes it beside what each charges a buyer for the same work. Most state neither, and that is recorded rather than estimated.
Why it costs you money
It does not change your bill, and it is the fact this market publishes and nobody collates. Where a service states both figures, the difference between what a solver earns and what a buyer pays is visible on that service's own site. It is the same question this site asks about where a residential IP address comes from: somebody is on the other end of the transaction, and the terms are checkable.
See it in the data
Related terms
- Captcha solving serviceA third party that clears a challenge for you and returns the token, billed per 1,000 solves rather than per gigabyte.
- Solve rateThe price for clearing 1,000 challenges of one type. The unit every captcha solving service quotes, and it differs per type.
- CAPTCHA rateHow often a target serves a challenge instead of the page. A softer signal than a ban and it costs the same bytes.
- Every term we define
Questions
Why does this belong on a pricing site?
Because both numbers come from the same company's own pages and neither is hard to check. This site already asks where a residential IP address comes from and whether the household agreed. The same question applies when a person, rather than a model, is doing the work you are buying.
Does every service use people?
No, and most do not say. Read 3 Oct 2026, some state that a model tries first and a person takes what it fails, one states that people do all of it, and the rest do not say either way on the pages we read. Not stated is not the same as automated.
Is the gap between the two figures a markup?
Proxy Compare does not calculate one, and the reason is that the two figures are not always measured the same way. A wage is usually quoted without naming a challenge type and a buyer rate always names one, so dividing them would rest on an assumption about which type was solved. Both figures are published here, sourced and dated, and the reader draws the conclusion.