Leaderboard / Compare

Compare AI humanizers

Select two to four AI humanizers to see their bypass rate, meaning preservation and readability in one table, or open any pair from the grid below. All figures on this page come from the November 2026 run. Every tool got the same prompts and was judged by the same detectors and scoring code. All of the data is public.

Pick 2 to 4 tools 3 of 4 selected

Copy linkReset

✓CleverHumanizer#1 · 79.33✓UndetectedGPT#2 · 76.03✓Rephrasy#3 · 72.97GPTinf#4 · 72.05SuperHumanizer#5 · 69.41SmartHumanizer#6 · 67.41HIX Bypass#7 · 60.04StealthGPT#8 · 52.87ReHumanize#9 · 44.65Undetectable AI#10 · 33.71

	CleverHumanizer#1 · 79.33×	UndetectedGPT#2 · 76.03×	Rephrasy#3 · 72.97×
Overall
Leaderboard rank	#1	#2	#3
Overall score	79.33	76.03	72.97
Score componentsOverall score weights: 42% bypass, 32% meaning, 16% readability, 10% consistency. Penalty points are then subtracted.
Bypass rate95% confidence interval shown below each value	89.682.7–96.5	70.759.6–81.8	80.070.1–89.8
Meaning preservation	63.7	79.0	69.6
Readability	79.5	83.7	75.3
Consistency across categories	86.2	76.5	80.7
Penalties	None	None	−3.0
Detector bypass ratesAverage human-likelihood (0–100) each detector gave this tool's outputs.
GPTZero	83.1	61.2	82.6
Originality.ai	31.1	7.7	28.1
Copyleaks	87.9	75.8	87.9
Pangram	8.8	6.2	17.2
ZeroGPT	81.1	71.3	75.5
QuillBot	90.4	92.3	83.9
Grammarly	96.9	93.1	86.5
Writing categoriesBypass, meaning preservation and readability on that category's prompts, weighted as in the overall score, plus the tool's overall consistency score. Penalties are not subtracted.
Academic essay6 prompts	81.0	82.2	78.4
Application essay6 prompts	68.1	67.7	61.4
Blog post6 prompts	81.2	69.3	79.0
Business email3 prompts	78.6	60.0	70.2
Marketing copy6 prompts	84.8	83.5	82.3
Discussion board3 prompts	79.8	83.5	84.7
News article3 prompts	84.0	87.4	78.7
AccessWhether the tool has a free tier. Every tool is tested on its default settings.
Free tier	Yes	Yes	—

Every matchup

Each cell shows how many points the row tool scores above (green) or below (red) the column tool in the November 2026 cycle. Select a cell to open the full comparison of the two. In the W–L column, W is the number of other tools that tool outscores and L is the rest.

45 pairs · 10 tools

Row minus column	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	W–L
#1CleverHumanizer79.33		+3.3	+6.4	+7.3	+9.9	+11.9	+19.3	+26.5	+34.7	+45.6	9–0
#2UndetectedGPT76.03	−3.3		+3.1	+4.0	+6.6	+8.6	+16.0	+23.2	+31.4	+42.3	8–1
#3Rephrasy72.97	−6.4	−3.1		+0.9	+3.6	+5.6	+12.9	+20.1	+28.3	+39.3	7–2
#4GPTinf72.05	−7.3	−4.0	−0.9		+2.6	+4.6	+12.0	+19.2	+27.4	+38.3	6–3
#5SuperHumanizer69.41	−9.9	−6.6	−3.6	−2.6		+2.0	+9.4	+16.5	+24.8	+35.7	5–4
#6SmartHumanizer67.41	−11.9	−8.6	−5.6	−4.6	−2.0		+7.4	+14.5	+22.8	+33.7	4–5
#7HIX Bypass60.04	−19.3	−16.0	−12.9	−12.0	−9.4	−7.4		+7.2	+15.4	+26.3	3–6
#8StealthGPT52.87	−26.5	−23.2	−20.1	−19.2	−16.5	−14.5	−7.2		+8.2	+19.2	2–7
#9ReHumanize44.65	−34.7	−31.4	−28.3	−27.4	−24.8	−22.8	−15.4	−8.2		+10.9	1–8
#10Undetectable AI33.71	−45.6	−42.3	−39.3	−38.3	−35.7	−33.7	−26.3	−19.2	−10.9		0–9

The ranking is rebuilt each cycle from the published input texts, rewrites and detector results. The latest run was on November 1, 2026. How scoring works