AI humanizer pricing compared: list price per 10,000 words, dated

The table carries the cheapest paid plan each vendor lists, its monthly list price where the page shows one, its yearly rate as the vendor shows it, and the two limits that decide what the money buys.

Sources: Undetectable AI pricing, StealthGPT pricing, HIX Bypass pricing and GPTHuman pricing, all read 9 November 2026; the remaining vendors' pages are cited on their review pages, read 8 and 9 November 2026. Walter Writes AI shows its pricing on the homepage with the yearly toggle on by default, so its monthly list price is one of the figures this page could not read; its main limitation for a buyer comparing on price is the 750-word request cap on the plan that carries the $8 rate.

Several pages default to the yearly toggle, and a card that reads "$5.00 a month" on such a page describes a $60.00 charge. The column "Charged for the year" is there for that reason, and it is blank where the page did not print the total.

Dividing each plan's price by its monthly allowance puts word-metered plans on one scale. The figure is computed by the editors from the list prices in the table above, read on 8 and 9 November 2026, and it is a list price per 10,000 words of allowance, not a measurement of anything the tool does.

Three readings follow, and none of them is a verdict. First, the spread among word-metered yearly rates is about seven to one, from $2.67 to $19.98, which is wider than the spread in headline prices suggests. Second, the monthly and yearly columns rank the same tools differently: Undetectable AI's $9.99 is the lowest monthly list price in the table, while its yearly per-10,000 figure sits fourth. Third, four of the twelve tools cannot be placed on the scale at all, and a buyer comparing them has to compare meters rather than prices.

WriteHuman's figure deserves its caveat. Basic sells 80 requests a month at up to 600 words each, read 9 November 2026, so a writer who sends 200-word paragraphs gets 16,000 words from the plan and pays $7.50 per 10,000 on the yearly rate, three times the ceiling figure. A request meter rewards long inputs and punishes short ones.

A monthly pool and a request cap are two limits on the same money, and the smaller one decides how a given document gets through.

Caps read 8 and 9 November 2026. Each request is rewritten without sight of the others, so a document split four ways has three joins to check for a repeated point or a term that changed halfway through. That cost does not appear on any pricing page, and it is the reason this site treats words per request as a limit worth scoring, at 15 points of 100 under the method's limits and access heading, while price is reported only.

Four questions sort most of it. Is the page on its yearly toggle, and what is the single charge? Is the plan metered in words, requests, prompts or rewrites a day? What is the request cap, and how many pieces does your usual document become? And does the price buy a humanizer alone or a suite, as at QuillBot and Grammarly, where the humanizer is one tool among several and the price cannot be attributed to it?

What a price does not tell you is whether the rewrite keeps your meaning. The method on this site weights meaning kept at 35 points of 100 and errors added at 20, rated by two editors blind against the input, and those figures appear on each review once a run is published. No detector score should be the only evidence in a misconduct case, and no price should be the only reason to pick a tool.

Refund windows and cancellation routes vary as much as the prices and are set out in AI humanizer refund policies; the tools' free allowances and paid trials are on the free humanizer page. Every tool in the table sits in the full humanizer ranking with a review on the same fields, and the guides hub carries the read date on each page so a figure here can be checked against the vendor's page on the day you read it.