ProofreaderPro.ai
AI Text Humanization

Humanization Intensity Settings Explained: Mild to ULTRA in ProofreaderPro

What each of the six humanization intensity settings in ProofreaderPro does, the AI score reduction measured at each one, and when ULTRA is the right choice.

Lisa - Author at ProofreaderPro.aiLisa|Nov 3, 2026|6 min read
humanization intensity settings - ProofreaderPro Blog

The humanization intensity setting decides how far a rewrite departs from your draft. ProofreaderPro's AI Humanizer has six settings on one slider: Mild, Mild+, Balanced, Strong, Aggressive and ULTRA. Each setting up rewrites more of every sentence and increases the AI score reduction, which in our tests on GPTZero goes from 50% at Mild to 89% at ULTRA.

ULTRA is the new sixth setting, included in Academic Plus. This guide covers what each setting changes in a text, what ULTRA costs in readability and grammar, how to revise an ULTRA output, and which setting suits a given document.

What the intensity setting controls

Intensity controls how much of each sentence the humanizer rewrites. At the low end it changes word choice and short phrases, so your sentence order and most of your vocabulary survive. At the high end it splits and merges sentences and changes the rhythm of whole paragraphs.

AI detectors read statistical patterns such as word predictability (perplexity) and sentence-length variation (burstiness). A deeper rewrite changes more of those patterns, which is why the AI score reduction grows with the setting. It also takes the text further from your draft, so semantic faithfulness decreases a little at each setting.

The six settings on the humanization intensity slider

SettingWhat it doesAI score reduction (GPTZero)
MildThe lightest rewrite, closest to your wording50%
Mild+A light rewrite, readability & grammar prioritized59%
BalancedThe recommended default68%
StrongA deeper rewrite, the AI score reduction prioritized76%
AggressiveA heavy rewrite, with a high AI score reduction78%
ULTRAThe maximum rewrite, the maximum AI score reduction possible89%

The graph below plots the AI score reduction on GPTZero, semantic faithfulness as NLI support, and grammar as LanguageTool flags per 1,000 words, at each of the six humanization intensities.

Line graph of the six intensity settings from Mild to ULTRA: the AI score reduction on GPTZero increases from 50% to 89%, semantic faithfulness decreases from 90.6 to 82, and LanguageTool flags decrease from 19.6 to 16.9 per 1,000 words

ULTRA: the deepest rewrite

ULTRA works on a fine-tuned model of its own, separate from the five settings below it. It rewrites more of every paragraph than any other setting, and its AI score reduction on GPTZero, 89% in our tests, is the highest from among the six points. Aggressive, the next setting down, measured 78%.

Readability and grammar are slightly lower, and semantic faithfulness is the lowest of the six settings. An ULTRA output needs a careful read before submission. The figures here are measurements on GPTZero at the time of this writing. Note that detector models change over time.

What ULTRA preserves

Some things are the same at every setting. Field-specific terms, in-text citations and reference entries are preserved verbatim, as are numbers, units and the terms you freeze. Headings, lists, tables and equations are returned as written. The post on an AI paraphraser that preserves citations covers how this works in each referencing style.

Try ULTRA in the AI Humanizer

Academic Plus includes the six-setting intensity slider with ULTRA, the Doc Humanizer, the voice filter and 50,000 words a month.

See Academic Plus

How to revise an ULTRA output

Always revise output before submission. A quick read and edit helps to ensure the content expresses your intended meaning. This can be done using the diff view in the humanizer dashboard.

  1. Open the Diff view and read the changes paragraph by paragraph, with your draft beside the output.
  2. Check every claim against your draft: the numbers, the hedges and the direction of each finding.
  3. Open the Cites view and compare the in-text citations with your draft. Each citation is on the sentence it supports.
  4. Fix any sentence that changed in meaning by hand, or humanize that passage again at Aggressive and keep the better version.
  5. Use 'Proofread in Editor' to open the result as a document and clear the remaining grammar errors with a Light Proofreading.

Which setting to choose

Start with Balanced. It suits most academic writing, and its output needs the lightest review. If a Balanced humanization still leaves an AI writing style behind, go up to Strong or Aggressive and humanize the text again. The guide on how to humanize AI text covers the writing modes, freeze terms and the review involved with each round.

DocumentSettingWhy
Thesis chapter or journal manuscriptBalancedReadability & grammar come first, and every claim is checked by a reader
Abstract you have already edited by handMild or Mild+A light touch keeps the text close to your wording
Essay that still reads machine-made at BalancedStrong or AggressiveA deeper rewrite changes more of the sentence rhythm
Text where the AI score reduction matters mostULTRAThe highest AI score reduction, with time set aside to revise

Use ULTRA when the AI score reduction matters most and you have time to revise the output. A long document at ULTRA takes slightly longer to review than the same document at Balanced.

ULTRA in the Doc Humanizer and in other languages

The same six-setting slider opens in the Doc Humanizer popup, so a whole .docx can be humanized at ULTRA. The body paragraphs are rewritten at that setting, and the headings, tables, figures and references are preserved as written. The post on how to humanize a whole Word document covers the upload process.

ULTRA works on English text only. The multilingual humanizer covers 60+ languages on its own engine, and its highest setting is Aggressive, so non-English text set to ULTRA is humanized at Aggressive. The AI Humanizer page covers the engine in full, including the writing modes and the voice filter.

Frequently asked questions

Q: What is the highest humanization intensity setting in ProofreaderPro?

ULTRA, the sixth setting on the slider. It is the maximum rewrite, with the highest AI score reduction (89% on GPTZero in our tests), at the cost of slightly lower readability & grammar.

Q: Which intensity setting should I use for a thesis?

Start with Balanced, the recommended default for academic writing. Go up to Strong or Aggressive if the output still reads machine-written. Use ULTRA for a chapter only when you can revise the output carefully prior to submission.

Q: Does ULTRA work in other languages?

ULTRA works on English text. In the 60+ languages of the multilingual humanizer, the highest setting is Aggressive.

Q: Is ULTRA on the Academic plan?

ULTRA is included in Academic Plus, along with the whole intensity slider. The Academic plan humanizes at Balanced only.

AI Humanizer

Six intensity settings from Mild to ULTRA, with citations, figures and freeze terms preserved verbatim at every setting. ULTRA is included in the Academic Plus plan.

Lisa - Author at ProofreaderPro.ai
LisaCMO

Lisa holds a PhD in linguistics from NYU, and has always been curious about how computers can leverage applied linguistics to understand and communicate in human language and assist with editing human written content. She is currently the Chief Marketing Officer at ProofreaderPro, where she leads marketing across copy, social media, email, and offline channels.

Keep Reading

Try Text Humanizer Free

Join researchers from 200+ universities worldwide. Free to start, no credit card required.

Get Started Free
Proofreader Pro AI
Refine your research with ProofreaderPro.ai, the world's leading AI-powered proofreader, tailored for academic text.
ProofreaderProAI, Greenleaf Ave, Staten Island, 10310 New York
Ā© 2026 ProofreaderPro.ai. A leading academic proofreader, editor & humanizer. Made with ā¤ļø and linguistically sound syntax 🌳s