IntellectuaLead
English

Independent style analysis

Stylistic LLM Writing benchmark

Compare frontier language models writing quality based on stylistic metrics (grammatical, syntactic, rethorical, lexical, prosody...) and with human writing as a reference. Lower AI evidence means the model output is stylistically closer to the human baseline used in this benchmark. Since the leaderboard is based on a stylistic analysis and not blind human preferences or machine-learning discrimination like AI detectors, you can understand why a model sounds more human than another.

52 compared models and baseline 8 writing genres 9 equally weighted signals

Detailed model comparison

Scores update automatically when a new benchmark is published.

Low AI Medium AI High AI
RankModelAI ScoreSentence length diversityDistribution breadth, quantiles and tailsGPT-2 perplexity similarityModel-like token predictabilityNoun contextualizationContext supplied around nounsHeavy grammarLLM-favored constructions per 1,000 lexical tokensPersonal tonePerspective signals per 100 wordsAI SlopRhetorical templates per 1,000 wordsAI WordingOverused wording per 1,000 wordsPunctuation richnessRare-sensitive variety across ten mark familiesSection variationIntro-body-ending change
#116.2/100Low AI0.744Low AISentence lengths are broadly distributed.0.0139Low AIToken choices are within the lowest evidence quarter.0.853Low AINouns are well contextualized with linking, helping, and qualifying language.13.3/1kLow AIFew of the directional dense-grammar signals are elevated.7.3%Low AIPersonal-tone density is near the human anchor.6.5/1kLow AIFew repeated rhetorical templates.5.4/1kLow AIFew overused AI wording matches.0.766Low AIPunctuation is richly distributed across common and less-common mark types.0.31Low AIThe document changes shape naturally.
#227.3/100Low AI0.711Medium AISentence lengths use several bands.0Low AIToken choices are within the lowest evidence quarter.0.759Low AINouns are well contextualized with linking, helping, and qualifying language.15.1/1kLow AIFew of the directional dense-grammar signals are elevated.4.9%High AIVery little visible perspective or uncertainty.11.1/1kMedium AISome repeated rhetorical templates.4.5/1kLow AIFew overused AI wording matches.0.726Low AIPunctuation is richly distributed across common and less-common mark types.0.221High AIVery little style change across the document.
#328.9/100Low AI0.708Medium AISentence lengths use several bands.0Low AIToken choices are within the lowest evidence quarter.0.8Low AINouns are well contextualized with linking, helping, and qualifying language.18.7/1kLow AIFew of the directional dense-grammar signals are elevated.5.9%High AIVery little visible perspective or uncertainty.11.5/1kMedium AISome repeated rhetorical templates.3.6/1kLow AIFew overused AI wording matches.0.668Medium AISome less-common punctuation types are represented.0.207High AIVery little style change across the document.
#429.3/100Low AI0.668High AISentence lengths are concentrated in a few bands.0.0013Low AIToken choices are within the lowest evidence quarter.0.808Low AINouns are well contextualized with linking, helping, and qualifying language.15.1/1kLow AIFew of the directional dense-grammar signals are elevated.5.8%High AIVery little visible perspective or uncertainty.5.5/1kLow AIFew repeated rhetorical templates.5.1/1kLow AIFew overused AI wording matches.0.56High AIPunctuation is concentrated in relatively few mark types.0.231Medium AISome section-to-section change.
#530.2/100Low AI0.672High AISentence lengths are concentrated in a few bands.0.001Low AIToken choices are within the lowest evidence quarter.0.608High AINouns receive little relational or qualifying context.14.1/1kLow AIFew of the directional dense-grammar signals are elevated.5.8%High AIVery little visible perspective or uncertainty.10.8/1kLow AIFew repeated rhetorical templates.7.5/1kLow AIFew overused AI wording matches.0.758Low AIPunctuation is richly distributed across common and less-common mark types.0.303Low AIThe document changes shape naturally.
#633.2/100Low AI0.755Low AISentence lengths are broadly distributed.0Low AIToken choices are within the lowest evidence quarter.0.779Low AINouns are well contextualized with linking, helping, and qualifying language.18.5/1kLow AIFew of the directional dense-grammar signals are elevated.5%High AIVery little visible perspective or uncertainty.13.2/1kMedium AISome repeated rhetorical templates.4.9/1kLow AIFew overused AI wording matches.0.596High AIPunctuation is concentrated in relatively few mark types.0.209High AIVery little style change across the document.
#733.8/100Low AI0.759Low AISentence lengths are broadly distributed.0.0002Low AIToken choices are within the lowest evidence quarter.0.657Medium AINoun contextualization is moderate.20.1/1kLow AIFew of the directional dense-grammar signals are elevated.4.8%High AIVery little visible perspective or uncertainty.9.7/1kLow AIFew repeated rhetorical templates.11.9/1kMedium AISome overused AI wording.0.734Low AIPunctuation is richly distributed across common and less-common mark types.0.196High AIVery little style change across the document.
#834.6/100Low AI0.701Medium AISentence lengths use several bands.0.0001Low AIToken choices are within the lowest evidence quarter.0.741Low AINouns are well contextualized with linking, helping, and qualifying language.17.8/1kLow AIFew of the directional dense-grammar signals are elevated.5.1%High AIVery little visible perspective or uncertainty.13.9/1kMedium AISome repeated rhetorical templates.5.1/1kLow AIFew overused AI wording matches.0.609High AIPunctuation is concentrated in relatively few mark types.0.249Medium AISome section-to-section change.
#935.7/100Medium AI0.604High AISentence lengths are concentrated in a few bands.0.0004Low AIToken choices are within the lowest evidence quarter.0.686Medium AINoun contextualization is moderate.19.8/1kLow AIFew of the directional dense-grammar signals are elevated.5.6%High AIVery little visible perspective or uncertainty.7.1/1kLow AIFew repeated rhetorical templates.7.8/1kLow AIFew overused AI wording matches.0.646Medium AISome less-common punctuation types are represented.0.218High AIVery little style change across the document.
#1036/100Medium AI0.672High AISentence lengths are concentrated in a few bands.0.0145Low AIToken choices are within the lowest evidence quarter.0.784Low AINouns are well contextualized with linking, helping, and qualifying language.19.4/1kLow AIFew of the directional dense-grammar signals are elevated.6.2%Medium AISome personal or subjective language.12.9/1kMedium AISome repeated rhetorical templates.8.5/1kLow AIFew overused AI wording matches.0.58High AIPunctuation is concentrated in relatively few mark types.0.213High AIVery little style change across the document.
#1136.1/100Medium AI0.616High AISentence lengths are concentrated in a few bands.0.0157Low AIToken choices are within the lowest evidence quarter.0.747Low AINouns are well contextualized with linking, helping, and qualifying language.17.7/1kLow AIFew of the directional dense-grammar signals are elevated.6.3%Medium AISome personal or subjective language.9.5/1kLow AIFew repeated rhetorical templates.8.2/1kLow AIFew overused AI wording matches.0.605High AIPunctuation is concentrated in relatively few mark types.0.217High AIVery little style change across the document.
#1236.4/100Medium AI0.6High AISentence lengths are concentrated in a few bands.0.0282Low AIToken choices are within the lowest evidence quarter.0.845Low AINouns are well contextualized with linking, helping, and qualifying language.14.6/1kLow AIFew of the directional dense-grammar signals are elevated.7.2%Low AIPersonal-tone density is near the human anchor.8.7/1kLow AIFew repeated rhetorical templates.3.8/1kLow AIFew overused AI wording matches.0.603High AIPunctuation is concentrated in relatively few mark types.0.179High AIVery little style change across the document.
#1336.9/100Medium AI0.727Low AISentence lengths are broadly distributed.0.0897Medium AIToken choices are between one quarter and one half of the interval.0.746Low AINouns are well contextualized with linking, helping, and qualifying language.22/1kLow AIFew of the directional dense-grammar signals are elevated.4.7%High AIVery little visible perspective or uncertainty.12.2/1kMedium AISome repeated rhetorical templates.8.4/1kLow AIFew overused AI wording matches.0.696Low AIPunctuation is richly distributed across common and less-common mark types.0.182High AIVery little style change across the document.
#1437/100Medium AI0.673High AISentence lengths are concentrated in a few bands.0.0284Low AIToken choices are within the lowest evidence quarter.0.691Medium AINoun contextualization is moderate.21.5/1kLow AIFew of the directional dense-grammar signals are elevated.5.3%High AIVery little visible perspective or uncertainty.14.3/1kHigh AIFrequent repeated rhetorical templates.10.6/1kMedium AISome overused AI wording.0.684Medium AISome less-common punctuation types are represented.0.244Medium AISome section-to-section change.
#1537.1/100Medium AI0.66High AISentence lengths are concentrated in a few bands.0.036Low AIToken choices are within the lowest evidence quarter.0.747Low AINouns are well contextualized with linking, helping, and qualifying language.21.4/1kLow AIFew of the directional dense-grammar signals are elevated.5.8%High AIVery little visible perspective or uncertainty.8.9/1kLow AIFew repeated rhetorical templates.10.9/1kMedium AISome overused AI wording.0.631Medium AISome less-common punctuation types are represented.0.203High AIVery little style change across the document.
#1637.5/100Medium AI0.634High AISentence lengths are concentrated in a few bands.0Low AIToken choices are within the lowest evidence quarter.0.677Medium AINoun contextualization is moderate.14.4/1kLow AIFew of the directional dense-grammar signals are elevated.6%Medium AISome personal or subjective language.13.6/1kMedium AISome repeated rhetorical templates.8.7/1kLow AIFew overused AI wording matches.0.718Low AIPunctuation is richly distributed across common and less-common mark types.0.202High AIVery little style change across the document.
#1737.6/100Medium AI0.667High AISentence lengths are concentrated in a few bands.0.0033Low AIToken choices are within the lowest evidence quarter.0.821Low AINouns are well contextualized with linking, helping, and qualifying language.22.6/1kLow AIFew of the directional dense-grammar signals are elevated.5.5%High AIVery little visible perspective or uncertainty.11.3/1kMedium AISome repeated rhetorical templates.9.7/1kMedium AISome overused AI wording.0.56High AIPunctuation is concentrated in relatively few mark types.0.199High AIVery little style change across the document.
#1837.7/100Medium AI0.725Low AISentence lengths are broadly distributed.0.0016Low AIToken choices are within the lowest evidence quarter.0.553High AINouns receive little relational or qualifying context.23.4/1kLow AIFew of the directional dense-grammar signals are elevated.4.3%High AIVery little visible perspective or uncertainty.9.6/1kLow AIFew repeated rhetorical templates.14.5/1kHigh AIFrequent overused AI wording.0.74Low AIPunctuation is richly distributed across common and less-common mark types.0.252Medium AISome section-to-section change.
#1937.9/100Medium AI0.707Medium AISentence lengths use several bands.0.0004Low AIToken choices are within the lowest evidence quarter.0.836Low AINouns are well contextualized with linking, helping, and qualifying language.22/1kLow AIFew of the directional dense-grammar signals are elevated.5.8%High AIVery little visible perspective or uncertainty.15/1kHigh AIFrequent repeated rhetorical templates.8.2/1kLow AIFew overused AI wording matches.0.531High AIPunctuation is concentrated in relatively few mark types.0.235Medium AISome section-to-section change.
#2038.4/100Medium AI0.683High AISentence lengths are concentrated in a few bands.0Low AIToken choices are within the lowest evidence quarter.0.71Medium AINoun contextualization is moderate.19.6/1kLow AIFew of the directional dense-grammar signals are elevated.5.4%High AIVery little visible perspective or uncertainty.15.5/1kHigh AIFrequent repeated rhetorical templates.8/1kLow AIFew overused AI wording matches.0.634Medium AISome less-common punctuation types are represented.0.23Medium AISome section-to-section change.
#2138.7/100Medium AI0.709Medium AISentence lengths use several bands.0.0087Low AIToken choices are within the lowest evidence quarter.0.49High AINouns receive little relational or qualifying context.22.6/1kLow AIFew of the directional dense-grammar signals are elevated.4.5%High AIVery little visible perspective or uncertainty.9.8/1kLow AIFew repeated rhetorical templates.18/1kHigh AIFrequent overused AI wording.0.733Low AIPunctuation is richly distributed across common and less-common mark types.0.247Medium AISome section-to-section change.
#2238.8/100Medium AI0.699Medium AISentence lengths use several bands.0.1031High AIToken choices are beyond the interval midpoint.0.737Low AINouns are well contextualized with linking, helping, and qualifying language.27.9/1kLow AIFew of the directional dense-grammar signals are elevated.4.6%High AIVery little visible perspective or uncertainty.9.3/1kLow AIFew repeated rhetorical templates.20.9/1kHigh AIFrequent overused AI wording.0.692Low AIPunctuation is richly distributed across common and less-common mark types.0.181High AIVery little style change across the document.
#2338.9/100Medium AI0.743Low AISentence lengths are broadly distributed.0.0104Low AIToken choices are within the lowest evidence quarter.0.534High AINouns receive little relational or qualifying context.24.4/1kLow AIFew of the directional dense-grammar signals are elevated.4%High AIVery little visible perspective or uncertainty.9.7/1kLow AIFew repeated rhetorical templates.14.2/1kHigh AIFrequent overused AI wording.0.663Medium AISome less-common punctuation types are represented.0.272Low AIThe document changes shape naturally.
#2439.1/100Medium AI0.635High AISentence lengths are concentrated in a few bands.0Low AIToken choices are within the lowest evidence quarter.0.787Low AINouns are well contextualized with linking, helping, and qualifying language.17.4/1kLow AIFew of the directional dense-grammar signals are elevated.5.9%High AIVery little visible perspective or uncertainty.10.1/1kLow AIFew repeated rhetorical templates.3.7/1kLow AIFew overused AI wording matches.0.528High AIPunctuation is concentrated in relatively few mark types.0.168High AIVery little style change across the document.
#2539.4/100Medium AI0.695High AISentence lengths are concentrated in a few bands.0.0021Low AIToken choices are within the lowest evidence quarter.0.567High AINouns receive little relational or qualifying context.20.3/1kLow AIFew of the directional dense-grammar signals are elevated.5.7%High AIVery little visible perspective or uncertainty.13.3/1kMedium AISome repeated rhetorical templates.10.7/1kMedium AISome overused AI wording.0.682Medium AISome less-common punctuation types are represented.0.23Medium AISome section-to-section change.
#2639.4/100Medium AI0.714Medium AISentence lengths use several bands.0.0011Low AIToken choices are within the lowest evidence quarter.0.651Medium AINoun contextualization is moderate.22.4/1kLow AIFew of the directional dense-grammar signals are elevated.5.3%High AIVery little visible perspective or uncertainty.17.6/1kHigh AIFrequent repeated rhetorical templates.10.8/1kMedium AISome overused AI wording.0.731Low AIPunctuation is richly distributed across common and less-common mark types.0.214High AIVery little style change across the document.
#2741.1/100Medium AI0.672High AISentence lengths are concentrated in a few bands.0.9996High AIToken choices are beyond the interval midpoint.0.737Low AINouns are well contextualized with linking, helping, and qualifying language.18.5/1kLow AIFew of the directional dense-grammar signals are elevated.5.7%High AIVery little visible perspective or uncertainty.5.6/1kLow AIFew repeated rhetorical templates.15.6/1kHigh AIFrequent overused AI wording.0.688Medium AISome less-common punctuation types are represented.0.293Low AIThe document changes shape naturally.
#2841.3/100Medium AI0.676High AISentence lengths are concentrated in a few bands.0.0002Low AIToken choices are within the lowest evidence quarter.0.551High AINouns receive little relational or qualifying context.26.8/1kLow AIFew of the directional dense-grammar signals are elevated.4.7%High AIVery little visible perspective or uncertainty.8.8/1kLow AIFew repeated rhetorical templates.27.9/1kHigh AIFrequent overused AI wording.0.726Low AIPunctuation is richly distributed across common and less-common mark types.0.259Medium AISome section-to-section change.
#2941.6/100Medium AI0.626High AISentence lengths are concentrated in a few bands.0.0295Low AIToken choices are within the lowest evidence quarter.0.766Low AINouns are well contextualized with linking, helping, and qualifying language.20.5/1kLow AIFew of the directional dense-grammar signals are elevated.6.1%Medium AISome personal or subjective language.9.5/1kLow AIFew repeated rhetorical templates.7.7/1kLow AIFew overused AI wording matches.0.566High AIPunctuation is concentrated in relatively few mark types.0.159High AIVery little style change across the document.
#3042.3/100Medium AI0.636High AISentence lengths are concentrated in a few bands.0.007Low AIToken choices are within the lowest evidence quarter.0.656Medium AINoun contextualization is moderate.24.8/1kLow AIFew of the directional dense-grammar signals are elevated.6.4%Medium AISome personal or subjective language.13.7/1kMedium AISome repeated rhetorical templates.15.7/1kHigh AIFrequent overused AI wording.0.651Medium AISome less-common punctuation types are represented.0.197High AIVery little style change across the document.
#3142.5/100Medium AI0.579High AISentence lengths are concentrated in a few bands.0.0051Low AIToken choices are within the lowest evidence quarter.0.638High AINouns receive little relational or qualifying context.21.1/1kLow AIFew of the directional dense-grammar signals are elevated.6.7%Low AIPersonal-tone density is near the human anchor.9.6/1kLow AIFew repeated rhetorical templates.6.9/1kLow AIFew overused AI wording matches.0.625Medium AISome less-common punctuation types are represented.0.213High AIVery little style change across the document.
#3243.2/100Medium AI0.557High AISentence lengths are concentrated in a few bands.0Low AIToken choices are within the lowest evidence quarter.0.616High AINouns receive little relational or qualifying context.22.3/1kLow AIFew of the directional dense-grammar signals are elevated.7%Low AIPersonal-tone density is near the human anchor.8.4/1kLow AIFew repeated rhetorical templates.7/1kLow AIFew overused AI wording matches.0.589High AIPunctuation is concentrated in relatively few mark types.0.181High AIVery little style change across the document.
#3343.7/100Medium AI0.615High AISentence lengths are concentrated in a few bands.0.0051Low AIToken choices are within the lowest evidence quarter.0.657Medium AINoun contextualization is moderate.30.3/1kLow AIFew of the directional dense-grammar signals are elevated.6.4%Medium AISome personal or subjective language.11.1/1kMedium AISome repeated rhetorical templates.13.9/1kHigh AIFrequent overused AI wording.0.636Medium AISome less-common punctuation types are represented.0.215High AIVery little style change across the document.
#3443.8/100Medium AI0.677High AISentence lengths are concentrated in a few bands.0.629High AIToken choices are beyond the interval midpoint.0.791Low AINouns are well contextualized with linking, helping, and qualifying language.23/1kLow AIFew of the directional dense-grammar signals are elevated.5.2%High AIVery little visible perspective or uncertainty.10.5/1kLow AIFew repeated rhetorical templates.17.5/1kHigh AIFrequent overused AI wording.0.694Low AIPunctuation is richly distributed across common and less-common mark types.0.209High AIVery little style change across the document.
#3543.9/100Medium AI0.669High AISentence lengths are concentrated in a few bands.0.0116Low AIToken choices are within the lowest evidence quarter.0.613High AINouns receive little relational or qualifying context.21.7/1kLow AIFew of the directional dense-grammar signals are elevated.4.7%High AIVery little visible perspective or uncertainty.10.2/1kLow AIFew repeated rhetorical templates.10.5/1kMedium AISome overused AI wording.0.563High AIPunctuation is concentrated in relatively few mark types.0.194High AIVery little style change across the document.
#3644/100Medium AI0.577High AISentence lengths are concentrated in a few bands.0.0001Low AIToken choices are within the lowest evidence quarter.0.608High AINouns receive little relational or qualifying context.24.5/1kLow AIFew of the directional dense-grammar signals are elevated.6.7%Low AIPersonal-tone density is near the human anchor.5.8/1kLow AIFew repeated rhetorical templates.5.5/1kLow AIFew overused AI wording matches.0.552High AIPunctuation is concentrated in relatively few mark types.0.194High AIVery little style change across the document.
#3744.2/100Medium AI0.662High AISentence lengths are concentrated in a few bands.0.02Low AIToken choices are within the lowest evidence quarter.0.719Medium AINoun contextualization is moderate.24.4/1kLow AIFew of the directional dense-grammar signals are elevated.5.1%High AIVery little visible perspective or uncertainty.14.1/1kHigh AIFrequent repeated rhetorical templates.10.8/1kMedium AISome overused AI wording.0.601High AIPunctuation is concentrated in relatively few mark types.0.175High AIVery little style change across the document.
#3844.4/100Medium AI0.658High AISentence lengths are concentrated in a few bands.0.0028Low AIToken choices are within the lowest evidence quarter.0.605High AINouns receive little relational or qualifying context.25.5/1kLow AIFew of the directional dense-grammar signals are elevated.4.8%High AIVery little visible perspective or uncertainty.12.7/1kMedium AISome repeated rhetorical templates.8.8/1kLow AIFew overused AI wording matches.0.659Medium AISome less-common punctuation types are represented.0.189High AIVery little style change across the document.
#3944.5/100Medium AI0.652High AISentence lengths are concentrated in a few bands.0.349High AIToken choices are beyond the interval midpoint.0.781Low AINouns are well contextualized with linking, helping, and qualifying language.23/1kLow AIFew of the directional dense-grammar signals are elevated.6%Medium AISome personal or subjective language.10.2/1kLow AIFew repeated rhetorical templates.20/1kHigh AIFrequent overused AI wording.0.647Medium AISome less-common punctuation types are represented.0.182High AIVery little style change across the document.
#4045/100Medium AI0.685High AISentence lengths are concentrated in a few bands.0.3445High AIToken choices are beyond the interval midpoint.0.777Low AINouns are well contextualized with linking, helping, and qualifying language.25.1/1kLow AIFew of the directional dense-grammar signals are elevated.5.3%High AIVery little visible perspective or uncertainty.11.9/1kMedium AISome repeated rhetorical templates.21.1/1kHigh AIFrequent overused AI wording.0.637Medium AISome less-common punctuation types are represented.0.217High AIVery little style change across the document.
#4145.1/100Medium AI0.654High AISentence lengths are concentrated in a few bands.0.2912High AIToken choices are beyond the interval midpoint.0.689Medium AINoun contextualization is moderate.24.6/1kLow AIFew of the directional dense-grammar signals are elevated.5%High AIVery little visible perspective or uncertainty.12.1/1kMedium AISome repeated rhetorical templates.20.9/1kHigh AIFrequent overused AI wording.0.702Low AIPunctuation is richly distributed across common and less-common mark types.0.216High AIVery little style change across the document.
#4246.1/100Medium AI0.722Medium AISentence lengths use several bands.0.6349High AIToken choices are beyond the interval midpoint.0.772Low AINouns are well contextualized with linking, helping, and qualifying language.23.2/1kLow AIFew of the directional dense-grammar signals are elevated.5.2%High AIVery little visible perspective or uncertainty.15.6/1kHigh AIFrequent repeated rhetorical templates.15.3/1kHigh AIFrequent overused AI wording.0.703Low AIPunctuation is richly distributed across common and less-common mark types.0.187High AIVery little style change across the document.
#4346.4/100Medium AI0.669High AISentence lengths are concentrated in a few bands.0Low AIToken choices are within the lowest evidence quarter.0.483High AINouns receive little relational or qualifying context.25.5/1kLow AIFew of the directional dense-grammar signals are elevated.4.5%High AIVery little visible perspective or uncertainty.10/1kLow AIFew repeated rhetorical templates.19.7/1kHigh AIFrequent overused AI wording.0.654Medium AISome less-common punctuation types are represented.0.226High AIVery little style change across the document.
#4446.7/100Medium AI0.661High AISentence lengths are concentrated in a few bands.0.1528High AIToken choices are beyond the interval midpoint.0.723Medium AINoun contextualization is moderate.23.8/1kLow AIFew of the directional dense-grammar signals are elevated.5.2%High AIVery little visible perspective or uncertainty.10.6/1kLow AIFew repeated rhetorical templates.22.9/1kHigh AIFrequent overused AI wording.0.609High AIPunctuation is concentrated in relatively few mark types.0.236Medium AISome section-to-section change.
#4547.3/100Medium AI0.713Medium AISentence lengths use several bands.0.0351Low AIToken choices are within the lowest evidence quarter.0.564High AINouns receive little relational or qualifying context.24.2/1kLow AIFew of the directional dense-grammar signals are elevated.5.2%High AIVery little visible perspective or uncertainty.16.4/1kHigh AIFrequent repeated rhetorical templates.12.3/1kMedium AISome overused AI wording.0.652Medium AISome less-common punctuation types are represented.0.271Low AIThe document changes shape naturally.
#4647.4/100Medium AI0.647High AISentence lengths are concentrated in a few bands.0.0007Low AIToken choices are within the lowest evidence quarter.0.614High AINouns receive little relational or qualifying context.24.7/1kLow AIFew of the directional dense-grammar signals are elevated.4.8%High AIVery little visible perspective or uncertainty.20.8/1kHigh AIFrequent repeated rhetorical templates.23.5/1kHigh AIFrequent overused AI wording.0.757Low AIPunctuation is richly distributed across common and less-common mark types.0.21High AIVery little style change across the document.
#4747.6/100Medium AI0.706Medium AISentence lengths use several bands.0.0128Low AIToken choices are within the lowest evidence quarter.0.597High AINouns receive little relational or qualifying context.33.7/1kLow AIFew of the directional dense-grammar signals are elevated.5.4%High AIVery little visible perspective or uncertainty.15/1kHigh AIFrequent repeated rhetorical templates.18.3/1kHigh AIFrequent overused AI wording.0.609High AIPunctuation is concentrated in relatively few mark types.0.191High AIVery little style change across the document.
#4848.8/100Medium AI0.679High AISentence lengths are concentrated in a few bands.0.2718High AIToken choices are beyond the interval midpoint.0.735Low AINouns are well contextualized with linking, helping, and qualifying language.22.6/1kLow AIFew of the directional dense-grammar signals are elevated.5.6%High AIVery little visible perspective or uncertainty.12.1/1kMedium AISome repeated rhetorical templates.26.8/1kHigh AIFrequent overused AI wording.0.726Low AIPunctuation is richly distributed across common and less-common mark types.0.152High AIVery little style change across the document.
#4949.1/100Medium AI0.569High AISentence lengths are concentrated in a few bands.0.024Low AIToken choices are within the lowest evidence quarter.0.629High AINouns receive little relational or qualifying context.20.4/1kLow AIFew of the directional dense-grammar signals are elevated.6.6%Medium AISome personal or subjective language.9.9/1kLow AIFew repeated rhetorical templates.9.7/1kMedium AISome overused AI wording.0.521High AIPunctuation is concentrated in relatively few mark types.0.152High AIVery little style change across the document.
#5051/100High AI0.642High AISentence lengths are concentrated in a few bands.0.0001Low AIToken choices are within the lowest evidence quarter.0.592High AINouns receive little relational or qualifying context.39/1kMedium AISome dense constructions are elevated.6%Medium AISome personal or subjective language.15.5/1kHigh AIFrequent repeated rhetorical templates.14.2/1kHigh AIFrequent overused AI wording.0.619Medium AISome less-common punctuation types are represented.0.22High AIVery little style change across the document.
#5151.9/100High AI0.663High AISentence lengths are concentrated in a few bands.0.9944High AIToken choices are beyond the interval midpoint.0.79Low AINouns are well contextualized with linking, helping, and qualifying language.18.5/1kLow AIFew of the directional dense-grammar signals are elevated.6.2%Medium AISome personal or subjective language.5.2/1kLow AIFew repeated rhetorical templates.16.4/1kHigh AIFrequent overused AI wording.0.441High AIPunctuation is concentrated in relatively few mark types.0.213High AIVery little style change across the document.
#5255.8/100High AI0.61High AISentence lengths are concentrated in a few bands.0.4131High AIToken choices are beyond the interval midpoint.0.588High AINouns receive little relational or qualifying context.21.8/1kLow AIFew of the directional dense-grammar signals are elevated.5.3%High AIVery little visible perspective or uncertainty.13.8/1kMedium AISome repeated rhetorical templates.26.4/1kHigh AIFrequent overused AI wording.0.645Medium AISome less-common punctuation types are represented.0.266Low AIThe document changes shape naturally.
Labels summarize low, medium or high stylistic evidence. They compare writing tendencies and do not prove authorship.

Benchmark methodology

Nine writing signals used in every AI Score

Every model is evaluated with the same nine stylistic signals and the same set of writing genres. The score summarizes those signals across each model's completed tasks.

Sentence length diversity

Looks at whether a text naturally mixes short, medium and long sentences, including the center and tails of the distribution. Broader, less repetitive pacing is treated as more human-like.

GPT-2 perplexity similarity

Compares token predictability with patterns associated with language-model writing. It is one probabilistic signal, not a standalone detector.

Noun contextualization

Divides adpositions, auxiliaries and adverbs by nouns. A higher value means noun-led ideas receive more relational, tense or qualifying context. The calibration gives no AI evidence at 0.80 and full evidence at 0.50.

Heavy grammar

Tracks dense constructions including participial clauses, that-subject clauses, and 2-4 word adjective/noun sequences ending in an abstract -ty/-ity, -tion/-sion, -ness, or -ment noun.

Personal tone

Looks for four stance families: self-mentions including I, we, you, and your; hedges; boosters; and attitude markers, plus direct questions. Modal auxiliaries and multiword phrases such as “in practice” or “in fact” are classified inside those families and are not counted twice. A value of 7.4% gives zero AI evidence and 4.6% gives full AI evidence.

AI Slop

Tracks rhetorical templates such as em dashes, three-part sequences, formulaic contrasts, while-led concessions, generic setups, unsupported attribution and recap endings. Occasional use is normal; repetition is more informative.

AI Wording

Tracks words and short phrases that appear unusually often in LLM output. The metric is most useful when several generic matches accumulate in one text.

Section variation

Checks whether the opening, body and ending perform differently or repeat the same stylistic shape. More purposeful change is treated as more human-like.

Punctuation richness

Uses rare-sensitive Hill diversity at q = 0.25 across ten punctuation families. It rewards meaningful representation of less-common marks more strongly than ordinary entropy and contributes equally to the AI Score.

Controlled comparison

Eight writing genres

Using different genres reduces the chance that one topic or writing format determines the leaderboard.

  1. SEO article about coffee roasting
  2. Academic essay on universal basic income
  3. Research writing about AI detection
  4. Personal writing about overcoming insecurity
  5. Fiction about a Halloween party going wrong
  6. Biography of Alan Turing
  7. Copywriting email for a dating service
  8. News report on New York's cost of living

Questions

About the Stylistic LLM Writing benchmark

What does the AI Score measure?

It summarizes eight interpretable stylistic signals across the same controlled writing tasks. A lower result means the model's outputs are stylistically closer to the human reference used here; it does not mean every individual passage will sound human.

Does a high score prove that a text was written by AI?

No. The score compares measurable stylistic tendencies. It is useful for controlled model comparison, but it is not proof of authorship for an isolated document.

Why is the human baseline included?

The human samples provide a consistent reference under the same eight genres and tasks. They make the direction and scale of each model result easier to interpret without claiming that one human style is universally ideal.

How is this different from a preference leaderboard?

Preference leaderboards report which answer readers or judges favor. This benchmark measures explainable properties of the writing itself, including grammar, syntax, rhetorical habits, wording, sentence pacing and section structure.

How is this different from an AI detector?

AI detectors usually estimate whether an isolated text belongs to an AI or human class. This benchmark compares models under controlled prompts and exposes the stylistic signals behind the ranking. It should not be used to accuse an author or verify authorship.

Why are several genres used?

Writing style changes with purpose. Testing SEO, academic, research, personal, fiction, biography, email and news writing reduces the chance that one topic or format decides the ranking.

Why can a model's ranking change?

A ranking may change after new task outputs are published, a provider updates a model, missing analyses are completed, or the benchmark's linguistic analysis is improved. The table always uses the latest published runs.

Can two models with similar AI Scores still write differently?

Yes. Their overall scores can be close while their metric profiles differ substantially. Open the model details to compare which model relies more on dense grammar, formulaic rhetoric, predictable wording or uniform section structure.

How often does the leaderboard update?

The website reads the latest benchmark published from the Intellectualead app. Completed NLP backfills and new benchmark runs publish refreshed metrics automatically.