Humanizer · CMF vs Rephrasy
Pick two to four entries and inspect them side by side. All numbers come from the same shared source set and measurement profiles. An asterisk marks detector-dependent, provisional values.
| November 2026 | Humanizer · CMFVerified by HumanizeBenchWe ran this model ourselves on the benchmark source examples and checked its execution records. Verification confirms the tested model’s provenance; it does not guarantee performance or imply that every detector returned a result.infosave/humanizer-cmf Open weights | Closed source |
|---|---|---|
| Shared score components | ||
| Overall score | 45.23 | 55.34 |
| Panel bypass score | 30.2% | 78.0% |
| Meaning preservation | 65.9% | 35.7% |
| Raw meaning cosine | 0.9147 | 0.8328 |
| Readability | 78.0% | 73.4% |
| Category consistency | 79.9% | 78.5% |
| Detector coverage | 231/231 | 231/231 |
| Identical outputs | 0/33 | 0/33 |
| Penalty points | −3 | −12 |
| Per-detector human-likelihood scores · measured outputs only | ||
| GPTZero | 1.5% 33/33 | 49.7% 33/33 |
| ZeroGPT | 15.6% 33/33 | 51.9% 33/33 |
| Copyleaks | 42.4% 33/33 | 87.9% 33/33 |
| Originality.ai | 9.8% 33/33 | 34.9% 33/33 |
| Pangram | 30.8% 33/33 | 67.3% 33/33 |
| Grammarly | 69.5% 33/33 | 86.5% 33/33 |
| QuillBot | 99.6% 33/33 | 99.8% 33/33 |
| Category scores · before entry-level penalties | ||
| Academic essay | 43.3 | 70.0 |
| Application essay | 35.5 | 51.0 |
| Blog post | 54.6 | 74.2 |
| Business email | 35.3 | 55.0 |
| Marketing copy | 60.6 | 77.1 |
| Discussion board | 60.1 | 66.7 |
| News article | 47.0 | 74.5 |
| Evidence | All 33 outputs | All 33 outputs |
Compare the evidence, not just the total
Humanizer · CMF has 78.0 readability and Rephrasy has 73.4 under the shared rubric. Their raw meaning cosines are 0.9147 and 0.8328 respectively. Both entries have complete detector coverage.
Permanent comparison