November 2026: 15 services · five detectors · updated scoring

November 2026 · 750 outputs · 15 services

GPTZero detector benchmark

Model: 5.0b / 2026-11-04-base. Weight in the humanization composite: 30%.

predicted_class=human; mixed and ai fail. Human score is a class probability, not the percentage of human-written words.

November 2026 · methodology v2 · provisional automated quality. 15 humanizers, 50 identical inputs each, five detectors and eight quality metrics. Read the scoring rules and post-run changes.

A substantial second signal, informed by independent robustness and false-positive research; current-version shared calibration remains undone. Reference. None of these weights is a measured current detector accuracy.

GPTZero: ranked by strict-human pass rate, with raw averages separate.
HumanizerPasses/50Pass rateHuman-score average
Phrasly Ultra50/50100.00%99.31
WriteHuman40/5080.00%75.52
StealthGPT Super30/5060.00%52.21
Rephrasy22/5044.00%38.24
Undetectable.ai16/5032.00%26.50
ZeroGPT Humanizer15/5030.00%25.45
Clever Humanizer13/5026.00%21.86
Humbot5/5010.00%8.51
BypassGPT4/508.00%5.44
Grubby4/508.00%6.87
Grammarly Pro0/500.00%0.00
Solvely0/500.00%0.00
RewriteAI0/500.00%0.00
QuillBot0/500.00%0.00
HIX Bypass0/500.00%0.00

Failed strict bypass

100 minus the pass rate on these humanized outputs; this is not a false-positive rate on human writing.

GPTZero: failed strict bypass by service
HumanizerFailed/50Failed rate
Phrasly Ultra0/500.00%
StealthGPT Super20/5040.00%
WriteHuman10/5020.00%
Rephrasy28/5056.00%
Undetectable.ai34/5068.00%
Clever Humanizer37/5074.00%
ZeroGPT Humanizer35/5070.00%
BypassGPT46/5092.00%
Grubby46/5092.00%
Grammarly Pro50/50100.00%
Humbot45/5090.00%
Solvely50/50100.00%
RewriteAI50/50100.00%
QuillBot50/50100.00%
HIX Bypass50/50100.00%

This measures bypass of the tested detector, not its false-positive rate or accuracy on independently verified human writing. A high pass rate alone cannot establish that the detector is bad.

Per-output scores CSV · Frozen inputs and prompts CSV · Scoring manifest · Verification bundle · GitHub report · Earlier editions