HumanizerEval

Leaderboard / Compare / StealthGPT vs Rephrasy

September 2026 review · same 50 texts

StealthGPT Super vs Rephrasy

In the September 2026 HumanizerEval review, StealthGPT Super and Rephrasy rewrote the same 50 English texts. StealthGPT Super cleared all 5 detectors on 46 of them; Rephrasy on 20. The difference is mostly Pangram: on 24 samples StealthGPT passed it where Rephrasy failed, and the reverse happened on 1. On ZeroGPT and Winston AI the two were close to even.

Per-sample agreement

Per-sample agreement, September 2026. Counts are out of the 50 shared samples, computed at the 50% human-score threshold. Margin counts the samples only one tool passed, higher count first.
DetectorPassed more oftenMarginBoth passedNeither passed
Pangram 4.0StealthGPT24 to 1223
GPTZero 2026-09-13-baseStealthGPT6 to 1430
Originality.ai turboStealthGPT7 to 1420
ZeroGPT defaultRephrasy1 to 0490
Winston AI 5.0Rephrasy1 to 0490
All 5 detectorsStealthGPT26 to 0204

Every count is computed from scores.csv at build time, not transcribed. “Neither passed” counts the texts that defeated both tools; on Pangram that was 3 of 50 here.

Quality and output discipline

Quality and output discipline, September 2026.
MeasureStealthGPT SuperRephrasy
68%36%
72%30%
98%92%
96%60%
1.041.13
03
00
Super, production outputv4
Websitestealthgpt.ai rephrasy.ai

On per-sample quality scores, StealthGPT scored higher on 37 samples, Rephrasy on 1, and the two tied on 12. StealthGPT’s output was shorter on 32 of 50 samples; mean length ratios were 1.04 and 1.13, or 4% longer than the input and 13% longer than the input respectively.

Where StealthGPT outperforms Rephrasy

StealthGPT

  • Pangram: passed 46 of 50 samples, against 23 for Rephrasy.
  • GPTZero: passed 49 of 50 samples, against 44 for Rephrasy.
  • Originality.ai: passed 49 of 50 samples, against 43 for Rephrasy.
  • Factual consistency: kept facts accurate in 68% of outputs, against 36% for Rephrasy.
  • Naturalness: read fluently in 72% of outputs, against 30% for Rephrasy.
  • Syntax integrity: stayed grammatically clean in 98% of outputs, against 92% for Rephrasy.
  • Stance preservation: preserved the source’s stance in 96% of outputs, against 60% for Rephrasy.
  • Length: shorter than Rephrasy on 32 of 50 samples, which matters under a word limit.

Rephrasy

  • Pangram: failed 27 of 50 samples, 24 of which StealthGPT passed.
  • GPTZero: failed 6 of 50 samples, 6 of which StealthGPT passed.
  • Originality.ai: failed 7 of 50 samples, 7 of which StealthGPT passed.
  • Factual consistency: introduced factual errors in 64% of outputs.
  • Naturalness: had awkward or clunky phrasing in 70% of outputs.
  • Syntax integrity: had grammar or punctuation errors in 8% of outputs.
  • Stance preservation: distorted the source’s stance in 40% of outputs.
  • Length: longer than StealthGPT on 32 of 50 samples, so more to cut under a word limit.

StealthGPT vs Rephrasy FAQ

Is StealthGPT better than Rephrasy?

Yes. In the September 2026 HumanizerEval review, StealthGPT Super cleared all 5 detectors on 46 of the 50 shared samples and Rephrasy on 20. On ZeroGPT and Winston AI the two were within 6 points of each other. Quality ran 37 samples to 1, with 12 ties.

Which humanizer is best for Pangram, StealthGPT or Rephrasy?

StealthGPT Super. In September 2026, StealthGPT Super passed Pangram on 46 of the 50 shared samples and Rephrasy on 23. On 24 samples StealthGPT passed where Rephrasy failed; the reverse happened on 1. Both failed on 3.

Which humanizer changes the text less?

StealthGPT Super. It scored higher on per-sample quality on 37 of 50 samples, against 1. StealthGPT Super’s outputs averaged 1.04× the input length and passed factual consistency on 68%; Rephrasy’s averaged 1.13× and passed on 36%. On stance preservation the split was 96% to 60%. StealthGPT produced the shorter output on 32 of 50 samples.

More comparisons