Research report
Does Turnitin Detect Essays Generated by GPT-5.6-Sol?
A Benchmark Experiment on Turnitin AI Detection Accuracy for GPT-5.6-Sol Generated Essays
Published July 29, 2026
Overview
This report presents a benchmark experiment evaluating the accuracy of Turnitin's AI detection system when applied to essays generated by the GPT-5.6-Sol language model. The study uses a corpus of 180 essays (156,955 words) across multiple academic domains, levels, and word counts. Word-level accuracy is the sole metric reported.
- Overall word accuracy: 97.88% (153,620 correct words out of 156,955 total words).
- Domain performance:
- Education: 99.50% (5,003/5,028)
- Business: 99.32% (26,029/26,206)
- Humanities: 99.00% (41,865/42,288)
- STEM: 96.77% (75,429/77,946)
- Psychology: 96.48% (5,294/5,487)
- Major performance:
- Highest: Nursing (99.55%), Economics (99.54%), Criminal Justice (99.53%)
- Lowest: Physics (88.81%), Mathematics (89.57%), Computer Science (94.95%)
- Level performance:
- Graduate: 98.74% (77,902/78,899)
- Undergraduate: 97.00% (75,718/78,056)
- Word count performance:
- 400-word essays: 98.31% (23,849/24,259)
- 800-word essays: 98.42% (49,680/50,478)
- 1200-word essays: 97.41% (80,091/82,218)
- Dataset: GPT-5.6-Sol Generated Essay Corpus (156,955 words, 180 essays).
- Essay characteristics:
- Written at graduate and undergraduate levels.
- Cover 5 domains: Business, Education, Humanities, Psychology, STEM.
- Span 30 distinct majors (e.g., Accounting, Biology, Computer Science).
- Word counts: 400, 800, and 1200 words per essay.
- All essays are non-ESL and belong to the essay genre.
- Detection tool: Turnitin AI detection system.
- Metric: Word-level accuracy (percentage of words correctly flagged as AI-generated).
- Scoring unit: accuracy is measured in words. First identify correct paragraphs under the expectation below, then sum the words in those paragraphs for the correct-word count; total words are summed over all counted paragraphs.
- Correct paragraph: a counted paragraph that was flagged as AI. Correct words are the words in those paragraphs.
- References excluded: from a References/Bibliography-style heading onward, those paragraphs are left out of word counts.
- Domain and major coverage: Some domains (e.g., Education, Psychology) have only 6 essays each, limiting statistical power.
- Word count distribution: The 1200-word bucket has 61 essays, while 400 and 800 have 60 and 59 respectively, causing slight imbalance.
- Total essays: 180
- Total words: 156,955
- Correct words: 153,620
- Word accuracy: 97.88%
- Business: 99.32% (26,029/26,206)
- Education: 99.50% (5,003/5,028)
- Humanities: 99.00% (41,865/42,288)
- Psychology: 96.48% (5,294/5,487)
- STEM: 96.77% (75,429/77,946)
- Top 3 majors:
- Nursing: 99.55% (5,268/5,292)
- Economics: 99.54% (5,370/5,395)
- Criminal Justice: 99.53% (5,084/5,108)
- Bottom 3 majors:
- Physics: 88.81% (4,653/5,239)
- Mathematics: 89.57% (4,579/5,112)
- Computer Science: 94.95% (4,773/5,027)
- Graduate: 98.74% (77,902/78,899)
- Undergraduate: 97.00% (75,718/78,056)
- 400 words: 98.31% (23,849/24,259)
- 800 words: 98.42% (49,680/50,478)
- 1200 words: 97.41% (80,091/82,218)
Detection report screenshots
Turnitin marks suspected AIGC text in cyan.