https://humanizereval.com/ AI humanizer benchmark results against Pangram, GPTZero, Originality.ai, ZeroGPT and Winston AI. One item per review. en-us 1440 https://humanizereval.com/cycles/2026-09/ https://humanizereval.com/cycles/2026-09/ Wed, 30 Sep 2026 12:00:00 GMT StealthGPT Super leads at 96.8% mean strict bypass. StealthGPT Super: Pangram 4.0 46/50, GPTZero 2026-09-13-base 49/50, Originality.ai turbo 49/50, ZeroGPT default 49/50, Winston AI 5.0 49/50. WriteHuman: Pangram 4.0 32/49, GPTZero 2026-09-13-base 49/49, Originality.ai turbo 48/49, ZeroGPT default 49/49, Winston AI 5.0 49/49. Grubby: Pangram 4.0 18/50, GPTZero 2026-09-13-base 47/50, Originality.ai turbo 49/50, ZeroGPT default 50/50, Winston AI 5.0 50/50. Rephrasy: Pangram 4.0 23/50, GPTZero 2026-09-13-base 44/50, Originality.ai turbo 43/50, ZeroGPT default 50/50, Winston AI 5.0 50/50. Undetectable.ai: Pangram 4.0 16/50, GPTZero 2026-09-13-base 48/50, Originality.ai turbo 47/50, ZeroGPT default 50/50, Winston AI 5.0 49/50. RewriteAI: Pangram 4.0 14/50, GPTZero 2026-09-13-base 43/50, Originality.ai turbo 44/50, ZeroGPT default 44/50, Winston AI 5.0 48/50. HIX Bypass: Pangram 4.0 1/50, GPTZero 2026-09-13-base 34/50, Originality.ai turbo 32/50, ZeroGPT default 50/50, Winston AI 5.0 49/50. Humbot: Pangram 4.0 0/50, GPTZero 2026-09-13-base 31/50, Originality.ai turbo 35/50, ZeroGPT default 50/50, Winston AI 5.0 50/50. BypassGPT: Pangram 4.0 0/50, GPTZero 2026-09-13-base 28/50, Originality.ai turbo 33/50, ZeroGPT default 50/50, Winston AI 5.0 50/50. Ryne: Pangram 4.0 11/50, GPTZero 2026-09-13-base 4/50, Originality.ai turbo 0/50, ZeroGPT default 31/50, Winston AI 5.0 13/50. Sample: 50 identical English texts. Methodology v1.0.