September 2026 Benchmark Results
StealthGPT
AI humanizer included in the September 2026 AI Humanizer Benchmark.
StealthGPT wasn't part of the October 2026 cycle. These are its September 2026 results. See the current leaderboard for the latest rankings.
Scorecard
By detector
How StealthGPT performs against each detector we test.
| Detector | Score | Pass rate |
|---|---|---|
| Turnitin | 87.0 | 87% |
| GPTZero | 90.3 | 90% |
| Originality.ai | 84.2 | 84% |
| Winston AI | 85.1 | 85% |
By writing category
Performance across the categories of writing we score separately.
| Category | Score |
|---|---|
| SEO content | 91.6 |
| Marketing copy | 90.4 |
| Students | 88.6 |
| Essays | 85.4 |
| Academic writing | 83.7 |
At a glance
StealthGPT ranks second of seven with an overall score of 81.14. It trails GPTHuman by 7.14 points and leads Humanize AI Pro by 1.62. GPTHuman remains the top-ranked AI humanizer in the September 2026 leaderboard at 88.28. StealthGPT's strongest core result is detector bypass at 86.2, followed by readability at 84.1 and consistency at 82.6.
Why it stands out
StealthGPT is the category leader for SEO content at 91.6 and marketing copy at 90.4, making it the strongest specialist among the top three. It also ranks second against GPTZero at 90.3 and Originality.ai at 84.2. Those results make it especially relevant when web or persuasive writing is the main use case.
The trade-off
Meaning preservation is its weakest core measure at 78.9, 12.5 points behind GPTHuman and 3.2 behind Humanize AI Pro. It also places third for essays and academic writing rather than matching its category-leading SEO and marketing results. The benchmark applies penalties for meaning drift, split-detector outcomes and rerun variance.
Bottom line
StealthGPT is the best-supported choice for SEO and marketing copy in this cycle, but it is not the strongest all-round humanizer. Choose it when those categories matter most, and review the output carefully when preserving precise meaning is more important than detector performance.
Where it performed best
The benchmark case for StealthGPT
StealthGPT performed best on commercial and web-writing categories while remaining the second-ranked tool overall.
Second overall
81.14 overall, with 86.2 detector bypass, 78.9 meaning, 84.1 readability and 82.6 consistency.
SEO category leader
Ranked #1 for SEO content with a category score of 91.6.
Marketing category leader
Ranked #1 for marketing copy with a category score of 90.4.
Strong detector showing
Ranked #2 against GPTZero (90.3) and Originality.ai (84.2), and #3 against Winston AI (85.1).
Measured limitations
Where the result needs context
StealthGPT's strongest bypass and commercial-writing results came with clear trade-offs.
Meaning preservation
Its 78.9 meaning score was its weakest core measure and trailed both GPTHuman and Humanize AI Pro.
Academic categories
It placed third for essays and academic writing rather than matching its #1 SEO and marketing results.
Applied penalties
Its reported score includes penalties for split-detector results, meaning drift and rerun variance.
Point-in-time evidence
The findings describe one fixed September 2026 cycle and cannot guarantee the outcome for an individual document.
Compare StealthGPT
Head-to-head benchmark pages
Compare its SEO and marketing strengths with the tools immediately above and below it.
Related research
See how performance changes under different tests
Arena Labs examines questions that the main leaderboard does not answer on its own.