AGP Picks
View all

GPTHuman launches Origin 5.0 with 92.6% detector pass rate

Oct. 29, 2026
GPTHuman launches Origin 5.0 with 92.6% detector pass rate
By AI, Created 14:00 UTC, Oct 29, 2026, AGP -

GPTHuman has launched Origin 5.0, a new English-language AI humanization model that the company says preserved meaning across 500 one-pass tests while passing all four external AI detectors in 92.6% of cases. The release matters because it targets a growing need for writing tools that sound more natural without changing the original message.

Why it matters: - Origin 5.0 is built to make AI-assisted writing read more naturally without changing the argument, intent or central meaning of the source text. - The launch targets a core problem in AI rewriting: making text less detectable as machine-generated without distorting facts or meaning. - GPTHuman positions the model as a specialized alternative to general writing tools that focus more broadly on grammar correction and paraphrasing.

What happened: - GPTHuman launched Origin 5.0, its new English-language AI humanization model. - The company says Origin 5.0 is designed to improve phrasing, rhythm, transitions and sentence structure while preserving substance. - The model was tested on 500 AI-written English texts across eight source models and 10 writing categories. - Each text was processed once, with no retries or hand-picked replacements.

The details: - Four external detection tools evaluated every output: ZeroGPT, Originality.ai, Copyleaks and QuillBot. - An LLM quality judge compared each output with its source for generation glitches, meaning preservation and factual preservation. - 92.6% of outputs passed all four external AI detectors, or 463 of 500. - 99.8% passed at least three detectors, or 499 of 500. - 100% preserved the original meaning, including the source’s main point, stance and conclusion. - 99.4% contained no generation glitches, or 497 of 500. - 93% preserved every key fact, or 465 of 500. - 92.4% passed every quality check, or 462 of 500. - Average similarity measured 92.74 out of 100, indicating the rewritten text stayed close to the source while changing phrasing and flow. - Origin 5.0 received average human scores of 99.49 on ZeroGPT, 93.02 on Originality.ai, 99.40 on Copyleaks and 99.43 on QuillBot. - Source samples averaged 318 words, while humanized versions averaged 352 words. - Average character count fell 5.8%, and characters per word dropped from 6.82 to 5.62. - The benchmark showed 37% of outputs stayed within 10% of the source word count, 60% became more than 10% longer and 3% became more than 10% shorter. - GPTHuman says Origin 5.0 can help refine stiff or repetitive phrasing, awkward sentence flow, mechanical academic language, grammar, clarity and drafts that need improvement without a complete rewrite.

Between the lines: - The benchmark design emphasizes both detector performance and content preservation, rather than detector results alone. - The longer average output suggests the model often swaps dense phrasing for shorter, more conversational language while keeping the message intact. - The high pass rates across multiple detectors suggest the product is aimed at users who want text that appears more human across different detection systems.

What's next: - Origin 5.0 is available now in GPTHuman. - Users can try Origin 5.0 in GPTHuman. - The complete row-level benchmark results are available in the public GPTHuman Humanizer Benchmark. - The benchmark dataset drew from DetectRL-X and the University of Chicago Booth DetectionAI dataset, covering material from Anthropic, DeepSeek, Google, OpenAI and Qwen across news, web content, academic writing, SEO content, blogs, Wikipedia-style text, résumés, restaurant reviews, novels and Amazon reviews.

The bottom line: - GPTHuman is betting that the next generation of AI humanizers will be judged on two things at once: detector performance and faithful meaning preservation.

Disclaimer: This article was produced by AGP Wire with the assistance of artificial intelligence based on original source content and has been refined to improve clarity, structure, and readability. This content is provided on an “as is” basis. While care has been taken in its preparation, it may contain inaccuracies or omissions, and readers should consult the original source and independently verify key information where appropriate. This content is for informational purposes only and does not constitute legal, financial, investment, or other professional advice.

Sign up for:

Illinois Business Review

The daily local news briefing you can trust. Every day. Subscribe now.

By signing up, you agree to our Terms & Conditions.

Share this page:

Advanced Search Options

Search for:

Search scope:

Type:

Search in:

Date range:

The last

Sort by:

Sign up for:

Illinois Business Review

The daily local news briefing you can trust. Every day. Subscribe now.

By signing up, you agree to our Terms & Conditions.