Leaderboard / Humanizers / WriteHuman

September 2026 review · rank 2 of 10

WriteHuman AI detector test results

In the September 2026 HumanizerEval review, WriteHuman ranked 2 of 10 with a mean strict-bypass rate of 92.7%, behind StealthGPT Super at 96.8%. It cleared all 5 detectors on 32 of 49 texts. Review our full results for the best AI humanizer on the leaderboard page.

WriteHuman results, September 2026

Website

writehuman.ai

Rank

2 of 10

Mean strict bypassThe average of the five detector pass rates: the share of texts each detector called human-written. The leaderboard is ranked by this number.

92.7%

Passed all 5 detectorsTexts that all five detectors called human-written.

32 of 49

Pangram 4.0 strict bypassHow many texts Pangram called human-written (a human score above 50%), with the mean and median human score it gave across all texts.

32 of 49 (65%), mean 63.6%, median 91.39%

GPTZero 2026-09-13-base strict bypassHow many texts GPTZero called human-written (a human score above 50%), with the mean and median human score it gave across all texts.

49 of 49 (100%), mean 100.0%, median 100%

Originality.ai turbo strict bypassHow many texts Originality.ai called human-written (a human score above 50%), with the mean and median human score it gave across all texts.

48 of 49 (98%), mean 96.1%, median 99.98%

ZeroGPT default strict bypassHow many texts ZeroGPT called human-written (a human score above 50%), with the mean and median human score it gave across all texts.

49 of 49 (100%), mean 94.8%, median 98.2%

Winston AI 5.0 strict bypassHow many texts Winston AI called human-written (a human score above 50%), with the mean and median human score it gave across all texts.

49 of 49 (100%), mean 99.8%, median 99.83%

Factual consistency passShare of outputs where key claims, numbers, citations, and named entities remain accurate and correctly attributed.

22%

Naturalness passShare of outputs where wording is fluent and free of awkward or clunky phrasing.

49%

Syntax integrity passShare of outputs where sentences are grammatically complete and free of disruptive run-ons or punctuation errors.

98%

Stance preservation passShare of outputs where the source's position, tone, and intent are not reversed or distorted.

65%

Mean length ratioOutput word count divided by input word count, averaged across texts. 1.00 is the same length; 1.10 is 10% longer.

1.02

Samples over 1.4× inputOutputs more than 1.4× the input’s word count.

0

Samples under 0.6× inputOutputs shorter than 0.6× the input’s word count.

0

Settings usedThe mode each tool ran on: its strongest advertised setting on the cheapest paid tier.

API default

Recompute any of this from scores.csv. Strict bypass means an overall human verdict on a human score above 50%. WriteHuman returned 49 of 50 outputs, so its figures are out of 49.

What the vendor claims, and what we measured

Claim

“98% Copyleaks, 97% Turnitin, 97% ZeroGPT, 96% GPTZero, 96% Grammarly, 96% Winston AI, 95% Originality, and 94% Pangram.” (writehuman.ai , retrieved October 1, 2026) Shown on the homepage as the human score per platform after WriteHuman processing.

Measured

Strict bypass in September 2026: Pangram 65%, GPTZero 100%, Originality.ai 98%, ZeroGPT 100%, Winston AI 100%.

Verdict

Not verified. In September 2026, strict bypass was ZeroGPT 100% against the 97% claimed, GPTZero 100% against the 96% claimed, Winston AI 100% against the 96% claimed, Originality.ai 98% against the 95% claimed and Pangram 65% against the 94% claimed. Copyleaks, Turnitin and Grammarly are not in this panel.

Review history

Head to head

Because every tool saw the same texts, these pages report per-sample agreement, how often each tool bypassed detection on the same input, rather than a difference of averages.

• WriteHuman vs StealthGPT

  WriteHuman cleared all 5 detectors on 1 sample where StealthGPT did not; the reverse happened on 14. Both cleared 31.

• WriteHuman vs Grubby

  WriteHuman cleared all 5 detectors on 17 samples where Grubby did not; the reverse happened on 2. Both cleared 15.

• WriteHuman vs Rephrasy

  WriteHuman cleared all 5 detectors on 14 samples where Rephrasy did not; the reverse happened on 1. Both cleared 18.

• WriteHuman vs Undetectable.ai

  WriteHuman cleared all 5 detectors on 18 samples where Undetectable.ai did not; the reverse never happened. Both cleared 14.

• WriteHuman vs RewriteAI

  WriteHuman cleared all 5 detectors on 21 samples where RewriteAI did not; the reverse happened on 1. Both cleared 11.

• WriteHuman vs HIX Bypass

  WriteHuman cleared all 5 detectors on 31 samples where HIX Bypass did not; the reverse never happened. Both cleared 1.

• WriteHuman vs Humbot

  WriteHuman cleared all 5 detectors on 32 samples where Humbot did not; the reverse never happened. Both cleared 0.

• WriteHuman vs BypassGPT

  WriteHuman cleared all 5 detectors on 32 samples where BypassGPT did not; the reverse never happened. Both cleared 0.

• WriteHuman vs Ryne

  WriteHuman cleared all 5 detectors on 32 samples where Ryne did not; the reverse never happened. Both cleared 0.