Leaderboard / Compare

Compare AI humanizers

Pick two to four tools and read them side by side, or jump to any pair from the matchup grid. Every number comes from the same September 2026 benchmark run: the same prompts, the same detectors, the same scoring, all of it published.

Pick 2 to 4 tools 3 of 4 selected

Copy linkReset

✓UndetectedGPT#1 · 84.90✓SmartHumanizer#2 · 84.80✓WriteHuman#3 · 81.60GPTinf#4 · 78.30AI Humanize#5 · 76.10HIX Bypass#6 · 73.80StealthGPT#7 · 71.90SuperHumanizer#8 · 68.50CleverHumanizer#9 · 66.80ReHumanize#10 · 39.00Undetectable AI#11 · 38.20

Any matchup cell or pair below jumps straight to that comparison.

	UndetectedGPT#1 · 84.90×	SmartHumanizer#2 · 84.80×	WriteHuman#3 · 81.60×
Overall
Leaderboard rank	#1	#2	#3
Overall score	84.90	84.80	81.60
Score componentsBypass 42%, meaning 32%, readability 16%, consistency 10%; penalties come off the total.
Bypass rate95% CI shown under each value	86.379.0–93.5	85.379.4–91.2	90.483.6–97.1
Meaning preservation	84.8	87.6	76.2
Readability	77.3	75.7	72.4
Consistency across categories	91.5	88.3	86.4
Penalties	None	None	−1.0
Detector bypass ratesShare of runs each detector classified as human-written.
GPTZero	80.0	71.0	56.5
Winston AI	54.4	51.0	82.1
Originality.ai	79.2	78.2	59.2
ZeroGPT	85.1	82.6	92.0
Grammarly	89.3	86.9	94.2
QuillBot	92.7	94.2	92.2
Copyleaks	21.2	15.2	63.6
Writing categoriesCategory score: this tool's composite-equivalent within that category alone.
Academic essay6 prompts	89.2	89.8	86.2
Application essay6 prompts	86.2	83.3	84.9
Blog post6 prompts	77.6	82.4	80.6
Business email3 prompts	79.7	75.6	69.5
Marketing copy6 prompts	85.8	89.2	86.2
Discussion board3 prompts	90.3	88.3	83.1
News article3 prompts	85.9	79.3	79.6
AccessAs tested: every tool runs on its default settings and base plan.
Free tier	Yes	Yes	Yes

Every matchup

Each cell is the row tool's overall score minus the column tool's, September 2026 cycle. Green means the row leads, red the column. Click a cell to load that pair above; the last column is each tool's win–loss record against the rest of the field.

55 pairs · 11 tools

Row minus column	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	#11	W–L
#1UndetectedGPT84.90		+0.1	+3.3	+6.6	+8.8	+11.1	+13.0	+16.4	+18.1	+45.9	+46.7	10–0
#2SmartHumanizer84.80	−0.1		+3.2	+6.5	+8.7	+11.0	+12.9	+16.3	+18.0	+45.8	+46.6	9–1
#3WriteHuman81.60	−3.3	−3.2		+3.3	+5.5	+7.8	+9.7	+13.1	+14.8	+42.6	+43.4	8–2
#4GPTinf78.30	−6.6	−6.5	−3.3		+2.2	+4.5	+6.4	+9.8	+11.5	+39.3	+40.1	7–3
#5AI Humanize76.10	−8.8	−8.7	−5.5	−2.2		+2.3	+4.2	+7.6	+9.3	+37.1	+37.9	6–4
#6HIX Bypass73.80	−11.1	−11.0	−7.8	−4.5	−2.3		+1.9	+5.3	+7.0	+34.8	+35.6	5–5
#7StealthGPT71.90	−13.0	−12.9	−9.7	−6.4	−4.2	−1.9		+3.4	+5.1	+32.9	+33.7	4–6
#8SuperHumanizer68.50	−16.4	−16.3	−13.1	−9.8	−7.6	−5.3	−3.4		+1.7	+29.5	+30.3	3–7
#9CleverHumanizer66.80	−18.1	−18.0	−14.8	−11.5	−9.3	−7.0	−5.1	−1.7		+27.8	+28.6	2–8
#10ReHumanize39.00	−45.9	−45.8	−42.6	−39.3	−37.1	−34.8	−32.9	−29.5	−27.8		+0.8	1–9
#11Undetectable AI38.20	−46.7	−46.6	−43.4	−40.1	−37.9	−35.6	−33.7	−30.3	−28.6	−0.8		0–10

Rankings are recomputed every cycle from published inputs, outputs and detector verdicts, last run September 3, 2026. How scoring works