September 2026 humanizer benchmark
How Phrasly Ultra compared on the same texts
In this benchmark, 4 AI humanizers rewrote the same 72 long AI-generated English texts. Pangram scored every output. Phrasly Ultra averaged a 99.8% human score, and 70 of 72 outputs scored 100% human.
September 2026 benchmark. 4 AI humanizers rewrote the same 72 texts. Every output was scored with Pangram, and outputs were checked for hallucinations.
Higher Pangram human scores and higher shares of 100% human and hallucination-free outputs are better. Fewer hallucinations per output is better. WriteHuman declined 3 of 72 texts. Declined texts returned no output and are excluded from all of that tool's averages. Hallucination counts were not recorded for 1 StealthGPT output and 2 Undetectable AI outputs. Those outputs are excluded from the hallucination measures only.