Best AI Humanizer for Pangram

Pangram catch rate

Caught means Pangram gave a human-likelihood score below 0.50 on a 0 to 1 scale. Avg human score is the mean score on a 0 to 100 scale. Rank 1 caught the most that month.

Pangram joined the benchmark in October 2026; earlier months were tested without it.

Humanizers that pass Pangram

Every AI humanizer we tested against Pangram in October 2026, ordered by its average Pangram score. 1 of 14 passed Pangram on at least 80% of its outputs. The top three passed 26, 22 and 15 of their 30 outputs.

Pangram humanizer results, October 2026

Rank Position by average Pangram score this cycle. Learn more →	Humanizer	ScorePangram score Average of Pangram's human-likelihood scores on this tool's outputs, where 100 means every output was scored as human. The table is ordered by this column. Learn more → ↓	Passed Outputs Pangram scored as human (a human-likelihood of 0.50 or higher), out of the tool's outputs it scored. Ranked by share passed. Click a count to read this tool's outputs. Learn more →	Overall Overall score across all detectors, after penalties. Learn more →
1	Phrasly phrasly.ai	85.6	26/30	61.79
2	StealthGPT stealthgpt.ai	69.9	22/30	74.50
3	WriteHuman writehuman.ai	50.3	15/30	74.89
4	Clever AI Humanizer cleverhumanizer.ai	26.5	8/30	69.02
5	Undetectable.ai undetectable.ai	3.6	1/30	71.18
6	Humanize AI Pro humanizeai.pro	0.3	0/30	68.87
7	AI Humanize io aihumanize.io	0.0	0/30	68.24
8	Super Humanizer superhumanizer.ai	0.0	0/30	67.58
9	Stealth Writer stealthwriter.ai	0.0	0/30	65.98
10	HIX Bypass bypass.hix.ai	0.0	0/30	65.14
11	Walter Writes walterwrites.ai	0.0	0/30	63.53
12	SupWriter supwriter.com	0.0	0/30	62.75
13	Humbot humbot.ai	0.0	0/30	61.74
14	Grammarly grammarly.com	0.0	0/30	61.29

Swipe the table sideways to see every column.

Gold, silver and bronze mark the top three in this month's columns, ties sharing a color; Passed is ranked by share of outputs passed.

Pangram compared with GPTZero and the other detectors

In October 2026, Pangram caught 82.9% of humanized outputs and GPTZero caught 36.0%. On the 420 outputs both scored this month, they gave the same verdict on 52.6%. Pangram gave the same verdict as Winston AI least often (23.3%).

The six detectors on the same humanized outputs, October 2026

Detector	Caught	Rate Catch rate	Avg score Avg human score	vs Sep 2026 Change since Sep 2026	Only it Only it caught	Agrees Same verdict as Pangram
Pangram	348 of 420	82.9%	16.9	n/a	147
GPTZero	151 of 420	36.0%	61.3	−11.2	1	52.6%
ZeroGPT	62 of 420	14.8%	76.5	−3.9	0	31.4%
Copyleaks	58 of 420	13.8%	85.9	−19.1	0	31.0%
Originality.ai	48 of 420	11.4%	87.2	−12.1	1	28.1%
Winston AI	30 of 420	7.1%	91.8	−28.3	1	23.3%

Swipe the table sideways to see every column.

Change is in points of catch rate. Only it caught counts outputs no other detector caught, of the 420 all six scored.

What Pangram catches

By writing type, Pangram caught humanized application essays most often (87%, 73 of 84) and news articles least often (76%, 32 of 42).

The second row pools the other detectors' verdicts on the same outputs. Each type had 3 to 6 prompts this month.

Tools that pass the other detectors but not Pangram

These humanizers passed the other five detectors at least 30 points more often than Pangram in October 2026, counting tools with 20 or more outputs Pangram scored.

Humanizers with a much lower pass rate on Pangram than on the other detectors, October 2026

Humanizer	Pangram Passed Pangram	Others Passed the others	Gap	Months Months with the gap
Humbot	0% 0/30	90% 135/150	−90 pts	1 of 1
Undetectable.ai	3% 1/30	93% 140/150	−90 pts	1 of 1
Walter Writes	0% 0/30	89% 133/150	−89 pts	1 of 1
Stealth Writer	0% 0/30	88% 132/150	−88 pts	1 of 1
AI Humanize io	0% 0/30	83% 124/150	−83 pts	1 of 1
Humanize AI Pro	0% 0/30	79% 118/150	−79 pts	1 of 1
Super Humanizer	0% 0/30	77% 115/150	−77 pts	1 of 1
HIX Bypass	0% 0/30	73% 109/150	−73 pts	1 of 1
Clever AI Humanizer	27% 8/30	93% 140/150	−67 pts	1 of 1
SupWriter	0% 0/30	64% 96/150	−64 pts	1 of 1
WriteHuman	50% 15/30	99% 148/150	−49 pts	1 of 1
Grammarly	0% 0/30	45% 68/150	−45 pts	1 of 1

Swipe the table sideways to see every column.

Outputs only Pangram caught

In October 2026, Pangram caught 147 humanized outputs that all five other detectors passed. That is 68% of the 215 outputs the others all let through. The three below have the widest gap between Pangram and the others, one per tool and prompt.

Raw scores run from 0 to 1, as on the evidence pages the prompt links open.

Of the 30 prompts, Pangram caught the largest share of tools on Cover Letter · gemini-3-8-flash (14 of 14) and the smallest share on four prompts, including Personal Statement · gemini-3-8-flash (10 of 14). All 30 October 2026 prompts

Choosing a humanizer for Pangram

The humanizer ranking above sorts by average Pangram score alone. Pangram does not give one verdict for a whole text: it labels each passage as human-written, AI-assisted, or AI-generated. The score here is the share of the text Pangram labelled human-written, so a passage it calls AI-assisted counts against the tool just as an AI-generated one does. A rewrite that only softens AI text into "AI-assisted" does not score well on this page. The methodology explains how each detector's result is turned into a score.

Pangram's current model, Pangram 4, was released in July 2026. It labels mixed and AI-assisted writing more finely than earlier versions and adds a check aimed at text rewritten to avoid detection, which is what humanizers do. Detectors change over time, so note which cycle you are reading and look at each tool's history on its results page.

A high Pangram score is not the whole picture. Check the Overall column in the ranking, and the Meaning and Readability columns on the leaderboard, before you choose, then try the tool on a paragraph of your own writing and read the output for changed facts and lost citations.

Questions about Pangram