AI humanizers compared directly
Rephrasy reaches a strong Human-plus-Mixed score. The complete comparison asks whether its rewrites also remain faithful, linguistically stable, and dependable with citations.
Internal benchmark · July 28, 2026. The same source texts, blinded review, and separate checks for meaning, language, length, and citations.
The result in one sentence
+33.2Rephrasy performed relatively well in the external AI check but lost substantial semantic fidelity and preserved no complete citation set exactly. Certentis delivered the more balanced output.
Overall quality
Quality scale 40–100 · truncated axis
Semantic fidelity
Exact citations
Distance from source quality
Best version in blinded review
Product profile and terms
The benchmark measures output quality. This section adds the current product context: modes, limits, languages, platforms, and publicly documented safeguards.
Certentis in everyday use
A controlled humanizer for academic and professional texts, with separate AI analysis and a traceable history.
Certentis does not optimize for a scanner classification alone. Automatic checks cover language, length, rewriting, and recognized protected content. Meaning is assessed in the archived benchmark and still requires your final review.
Product profile
A humanizer with multiple models and writing styles for individual users, plus an API for automated workflows.
Rephrasy V3 provides creative, journalistic, and professional style controls. In the archived test, its detector-oriented result was strong, but meaning and citations were substantially less stable.
| Features and terms compared | Certentis | Rephrasy v3 |
|---|---|---|
| Price and trial access | Credits; Humanizer uses 1 credit per input word. Internal retries within one process do not consume additional credits. | Free up to 100 words; Quick Pass $12.99. Subscriptions: Growth $18.99, Business $25, Professional $39 monthly; API from $0.10/1,000 words. |
| Input limit | Up to 30,000 characters and 3,000 words per input in the current web app. | Plain-text API up to 12,000 characters; structured HTML/Markdown API up to 200,000 characters at 2.5 times the billing rate. |
| Languages | Automatic language detection; the rewrite remains in the original's dominant language. | 50+ languages according to the homepage; English, German, Spanish, French, and Portuguese are highlighted for the strongest results. |
| Modes and workflow | One controlled mode with local checks followed by an analysis report. | V3, additional humanizer models, Creative, Journalistic, and Professional styles, plus cloned writing styles. |
| Academic texts and citations | Existing citation blocks are protected and checked for completeness after the rewrite. | No publicly documented citation lock, citation audit, or dedicated academic review workflow. |
| Platforms and API | Web app with history, AI analysis, and plagiarism checking in the same workspace. | Web, iOS, Android, Chrome extension, and APIs for the humanizer and detector. |
Public tutorials state different word limits. Unlimited offers also remain subject to a fair-use rule.
Official product sources
Provider information verified on August 3, 2026
Prices and features may change. Provider claims were not treated as evidence of quality; only the separate Certentis benchmark is used for that assessment.
The comparison at a glance
A single AI detector score cannot decide the result. A useful rewrite must also remain natural, faithful to the source, stable in language, and dependable in its citations.
| Metric | Certentis | Rephrasy v3 | Stronger result |
|---|---|---|---|
| Overall quality | 85.4 | 52.2 | Certentis |
| Semantic fidelity | 91.7 | 49.8 | Certentis |
| Exact citations | 8/10 | 0/10 | Certentis |
| Correct language | 10/10 | 10/10 | Tie |
| Word ratio | 0.98 | 1.08 | Certentis |
| Human + Mixed | 92.96% | 82.98% | Certentis |
Certentis: 10 usable cases · Rephrasy v3: 10 usable cases
Overall quality
The blinded review scored every output separately. These profiles show whether a tool merely sounds fluent or also preserves structure, precision, evidence, and meaning.
Rephrasy sometimes sounded fluent and academic, but lost the most ground in precision, citation integration, and semantic fidelity. The detector result concealed those differences.
| Metric | Certentis | Rephrasy v3 |
|---|---|---|
| Grammar & language | 93.9 | 77.6 |
| Clarity & readability | 91.8 | 65.5 |
| Coherence & structure | 89.4 | 56.0 |
| Academic style | 80.3 | 65.7 |
| Precision | 84.6 | 51.5 |
| Naturalness | 87.5 | 63.3 |
| Citation integration | 94.2 | 47.0 |
| Semantic fidelity | 91.7 | 49.8 |
Citeproc keys / Formatted citations
Five source texts used structured citeproc keys and five used already formatted references. That prevents format-specific weaknesses from disappearing inside an average.
Formatted citations were substantially more stable than citeproc. Even so, neither suite preserved a fully exact citation set, and citeproc fidelity fell to 37 points.
| Metric | Certentis | Rephrasy v3 |
|---|---|---|
| Quality | 85.0 | 41.0 |
| Fidelity | 90.6 | 37.0 |
| Citation integration | 92.8 | 38.0 |
| Citations exact | 3/5 | 0/5 |
| Human + Mixed | 97.4% | 85.7% |
| Word ratio | 0.96 | 1.28 |
| Metric | Certentis | Rephrasy v3 |
|---|---|---|
| Quality | 85.8 | 63.4 |
| Fidelity | 92.8 | 62.6 |
| Citation integration | 95.6 | 56.0 |
| Citations exact | 5/5 | 0/5 |
| Human + Mixed | 88.51% | 80.26% |
| Word ratio | 0.99 | 0.89 |
Observation from the benchmark
These observations come from individual cases in the archived run. They explain the metrics, but they are not blanket claims about every future output.
Rephrasy reached 82.98 percent Human + Mixed, illustrating that a strong detector result can coexist with weaker content.
In several citeproc cases, the review flagged literature, claims, or substantially different content that was absent from the source.
Some weak outputs sounded academic while central claims, footnotes, or complete sections were changed or removed.
What the numbers mean in practice
The right tool depends on whether you prioritize the highest detector classification, a particular writing style, or a controlled academic rewrite.
Certentis
detector orientation is acceptable only when factual, semantic, and citation fidelity remain intact.
Rephrasy v3
the external detector result is the priority and every claim and source will be checked against the original afterward.
Rephrasy is the clearest example of why a favorable detector score does not prove high text quality.
The 41.9-point semantic-fidelity gap signals material differences in facts, nuances, or relationships between claims.
Anyone publishing or submitting rewritten text should never read detector scores without comparing the meaning to the source.
Internal benchmark
Every provider received the same source texts. Outputs were scored against a fixed rubric without visible provider labels. Missing results, language drift, and changed citations remained visible.
Ten identical source texts with standard and structured citations
Blinded quality review across eight dimensions
Separate comparison of facts, qualifications, and intended meaning
Additional checks for language, word ratio, and complete citations
The results describe a dated run and do not guarantee every future output.
Provider models, prices, and product versions may change after the test.
Ten cases are a practical product test, not a complete representation of every text type.
Human + Mixed is an external document classification, not a percentage of human-written sentences.
Yes. Certentis is designed for users who want natural wording while retaining meaning and existing citations in a controlled workflow.
Certentis scored 92.96 percent and Rephrasy scored 82.98 percent.
It is an external document classification. It does not evaluate facts, citations, or fidelity to the original meaning.
No. This run is a dated sample of ten demanding multilingual texts. It reveals concrete strengths and risks, but it does not replace testing your own document.
The classification describes how an external scanner categorizes a document. It does not assess factual correctness, citation preservation, language stability, or argument quality.
Every reference present in the source had to remain complete and unchanged in the rewrite. A missing, reformatted, or damaged citation meant the case did not pass as exact.
Compare all seven tools
Every page uses the same benchmark run, so differences can be interpreted without changing the underlying dataset.
Overall quality: 82.4 · Semantic fidelity: 85.4
Read the full comparisonOverall quality: 70.0 · Semantic fidelity: 72.8
Read the full comparisonOverall quality: 42.7 · Semantic fidelity: 62.0
Read the full comparison