Almost every AI humanizer advertises a free tier. Very few mean the same thing by it, and the difference is not the word count. It is whether you get the actual output, or a preview with the useful part removed.
So this page is a checklist rather than a league table. Five things decide whether a free AI humanizer is usable, each one checkable in about a minute, and our own numbers are stated exactly so you can hold us to the same test. We are not publishing other vendors' free word caps here, for a reason given below.
In order of how much they matter, which is not the order vendors present them in.
1. Do you get the full output, or a preview? This is the one that matters and the one least often stated. Several tools rewrite your text, show you the first two sentences or a blurred block, and put the rest behind a signup or a card. The word "free" is doing a lot of work in those cases. Ours returns the whole rewrite, plus the score and the changed-word highlight, with nothing held back.
2. Does it need a card? A free tier that takes payment details is a trial. Both are legitimate; they are not the same product decision, and only one lets you evaluate without remembering to cancel. Ours takes no card and no email.
3. Does it reset, or expire? A one-off allowance of 500 words is spent in one sitting and then you are done. A daily allowance lets you come back, which matters because the useful test of a rewrite is whether it still reads as yours tomorrow. Ours resets daily: 3 humanizations a day, up to 260 words each.
4. Does it show you what changed? Almost none do, and it is the difference between a tool you can trust on real work and one you cannot. Without a diff you cannot tell whether a number moved, a claim softened, or a sentence appeared that you never wrote. Ours highlights the changed words.
5. Does it re-score its own output? A tool that rewrites and then measures what it produced is making a checkable statement. One that just asserts it worked is not. Ours scores before and after with our own detector, and labels the reading as ours.
Free tiers in this category change often, and we found our own published material disagreeing with itself about one competitor's free word cap while writing this. Rather than copy a figure we have not re-checked this week, the checklist above is the thing that stays true: run it against any tool you are considering, including ours. That is more useful than a table that is wrong by next month.
So you can run the five checks on us without taking anyone's word for it.
Anonymous, no account: 3 humanizations a day, up to 260 words per humanization. The 300 is a per-request ceiling rather than a daily pool, so it is three short rewrites rather than one long one. Resets daily. No card, no email.
What comes back each time: the full rewritten passage, an AI score before and after from our own detector with the reading labelled as ours, and a highlight of which words changed.
The detector separately: 3 scans a day at up to 1,000 words, also anonymous, on its own allowance rather than sharing the humanizer's.
What the free tier does not do: it does not take a file upload, so the humanizer box is pasted text only. It does not save anything, so there is nothing to come back to. And the allowance is keyed to your network rather than to you, which means on shared wifi other people can spend it. The mechanics of that are here.
Modes: Light varies sentence rhythm and cuts filler and leaves your vocabulary alone. Balanced edits phrasing more broadly. Maximum replaces most of the wording and is a paid mode, with the free tier covering Light and Balanced per our published plans. Current per-request sizes for each plan are on the plans page, not copied here, because a number on a comparison page goes stale and then misleads.
Three things, and the first one is the reason most people arrive here.
It cannot tell you whether you will pass a checker. Not ours, not any. Turnitin and the other major checkers do not expose their verdict to third-party tools, so a vendor publishing a pass rate is publishing a number it chose. We score with our own detector, which tells you about statistical properties of the text and is not a prediction of someone else's verdict.
It cannot make a rewrite safe to submit unread. This is the one that costs people. While testing for these pages, a Balanced pass returned a sentence that was not in our source at all: "Roughly a year into it." Nothing in the input mentioned any timescale. An invented detail that sounds specific is the real failure mode in this category, well ahead of style. Read the highlighted changes, especially around numbers, dates and citations.
It cannot defend you if you are already accused. If someone has flagged your work, do not rewrite the draft. Your version history is what defends you, and rewriting damages it. Gather evidence instead. And whether using one of these tools is permitted on your coursework is set by your institution, not by us.
Judge it on five things rather than on a ranking: whether you get the full output or a preview, whether it needs a card, whether the allowance resets or expires, whether it shows you what changed, and whether it re-scores its own output. Ours returns the full rewrite with no card and no account, 3 a day up to 260 words, resetting daily, with the changed words highlighted and a before and after score from our own detector. We disclose that this is our product, and the checklist is there so you can test us on it.
Ours takes no card and no email: 3 humanizations a day at up to 260 words, and you get the whole rewrite rather than a preview. The distinction worth watching across the category is between a free tier and a free trial. A tier that asks for payment details is a trial, which is a legitimate model but not one you can evaluate without remembering to cancel.
With us, 260 words per humanization and three humanizations a day, resetting daily. The 300 is a per-request ceiling rather than a daily total, so it is three separate short passages rather than one long one. We are not quoting other vendors free word caps on this page because those figures change often and we would be repeating something we have not re-checked.
Because the output is the thing they are selling, so showing you two sentences creates the incentive to sign up or pay. It is a reasonable business decision and it makes the tool impossible to evaluate, because the part you cannot see is the part you need to judge. If a tool blurs or truncates the rewrite, you are not testing it, you are being shown an advert for it.
The tool is not the risk; submitting its output unread is. Two things to actually worry about. A rewrite can invent a detail that was not in your source, which we observed in our own testing when a pass added a timescale nothing in the input mentioned. And if you are already under suspicion, rewriting the draft destroys the version history that would have defended you. Beyond that, whether the use itself is permitted is your institution decision rather than ours.
Not in the sense people expect. What a plan buys is size, persistence and your own allowance rather than a better rewrite of the same paragraph. The free tier covers Light and Balanced; Maximum is a paid mode and replaces most of the wording, which is heavier rather than automatically better, since the more wording a pass replaces the more chances it has to move a claim.
How the anonymous allowance is counted, and the one confusing failure.
Read the mechanics →What anonymous use cannot do, and how to evaluate in three rewrites.
See the trade →How to catch an invented detail before you submit it.
See the method →Longer passages, your own allowance, a saved history.
Compare plans →Three rewrites a day, 260 words each, full output, no card. The checklist above is there to be used against us too.