We Put Claude's Newest Model Up Against SEOZilla
Same titles. Same keywords. Same word counts. 33 articles SEOZilla shipped to real customers, re-written from scratch by Claude (Fable 5) — then both scored by ZeroGPT, the independent AI detector.
Benchmark run June 2026 · ZeroGPT "fakePercentage" · lower = more human
Raw Claude Fable 5
48.3%
average AI-detected across 33 articles
SEOZilla Humanised
24.2%
average AI-detected across the same 33 articles
No credit card required · Free site analysis included
How We Ran It (So You Can Check Our Work)
No tricks, no cherry-picking. We took the 33 most recent English-language articles our engine generated for real customer projects and gave Claude the exact same brief a normal person would type.
Same brief, both sides
For each article we reused the exact title, target keywords, and word count SEOZilla was given — the newest 33 English articles in a row, no skipping.
A fresh Claude, prompted like a normal user
Each draft came from a brand-new Claude session with zero extra instructions — no humanisation hints, no style tricks, first answer kept.
One judge: ZeroGPT
Both versions were scored through the identical ZeroGPT pipeline we use in production. Lower score = reads more human.
The exact prompt Claude got (example)
"Claude, I need a SEO article of about 1800 words, titled ‘…’, targeting these keywords: … Write it in markdown."
The Results
First the big picture, then every individual run. Yes, Claude beat us once — we published that too.
Where the 33 Articles Landed
Number of articles per ZeroGPT score band — SEOZilla clusters in the human zone, raw Claude Fable spreads deep into AI-detected territory.
0–20%
reads human
20–40%
mostly human
40–60%
mixed signals
60–80%
likely AI
80–100%
AI generated
Mean score
48.3%vs24.2%
Median score
47.3%vs22.4%
Every Single Run
All 33 head-to-head runs, sorted by SEOZilla's winning margin.
Bar length = ZeroGPT AI-detection score (0–100%, lower is better). Differences under 5 points are within detector noise and counted as ties.
The Scoreboard
27
SEOZilla wins
5
ties (within 5 pts)
1
Claude wins
The Stat That Should Worry You
Beyond the score, ZeroGPT issues a verdict on every article — and the "AI generated" verdicts mark exactly the kind of content Google and ChatGPT have been deranking. Each square below is one of the 33 articles, coloured by its actual recorded verdict.
Raw Claude Fable drafts
13 of 33
flagged as AI-generated by ZeroGPT — more than 1 in 3
SEOZilla articles
0 of 33
flagged as AI-generated. Zero. Not one article.
Claude Is a Brilliant Writer. That's Not the Problem.
We use frontier AI models inside SEOZilla too. The difference is what happens after the first draft: every article runs through our proprietary multi-agent humanisation gauntlet, verified against AI detectors before it ships.
A raw model, prompted once
- One draft, straight to you — no detector ever sees it
- Worst results on exactly the content blogs need: how-tos, guides, listicles
- Scored up to 95.9% AI-detected in this benchmark (a health how-to)
- Competitive only on newsy, analysis-style topics
SEOZilla's humanisation gauntlet
- A proprietary humanisation pipeline works over every single article
- Detector-verified before anything ships
- Never produced a hard "AI generated" verdict in this benchmark
- Plus real keyword research, internal links, images, and auto-publishing
The fine print (because benchmarks without it are marketing fiction)
- SEOZilla scores are the ones recorded when each article was generated; Claude drafts were scored fresh through the identical pipeline. We spot-checked that stored and fresh scores match.
- SEOZilla runs a proprietary humanisation pipeline on every article, while Claude got one raw shot — that head start is the product, not a flaw in the test.
- Single-article ZeroGPT scores wobble by a few points; we treat anything within 5 points as a tie.
- Claude model: claude-fable-5, June 2026. Detectors and models both evolve — we re-run benchmarks as they do.
- Why only 33 articles? We hit our Claude usage limits running the benchmark. More runs are coming — but we also need those tokens to ship features.
See What the Gauntlet Does to Your Content
Enter your website and get a free humanised article for your own niche — with the AI-detection score included, so you can verify it yourself.
No credit card required · Free site analysis included