Benchmarks

Academi.cx vs GPTZero

We benchmarked Academi.cx and GPTZero against Turnitin across 429 documents, measuring score differences, false flags, and missed AI.

Brayden · 2 min read

For this benchmark, we submitted the same 429 documents to Academi.cx and GPTZero through their APIs, then compared each tool’s scores with the documents’ actual Turnitin AI scores. Together those documents add up to close to 944,000 words, averaging about 2,200 words each, ranging from a 311 word short answer to a 25,000 word thesis.

How close were the predictions

Across all 429 documents, Academi.cx averaged 1.92 points off the real Turnitin score. GPTZero averaged 20.5 points off. Its median error sits at 3.77, meaning even the typical prediction missed by several points, well before factoring in its worst cases.

Paired scatterplots show the AI score for all 429 matched documents against Turnitin. Academi.cx appears in blue and GPTZero in gray, with identical 0 to 100 percent axes. The dashed diagonal marks matching scores. Mean absolute difference from Turnitin is 1.92 percentage points for Academi.cx and 20.50 for GPTZero. No documents are omitted; some points overlap.

Breaking it down by document type

The 429 documents split into 220 that Turnitin scored as 0% AI, 99 it scored as 100% AI, and 110 that fell somewhere in between.

Grouped bars show average absolute deviation from Turnitin in percentage points. All documents (n = 429): Academi.cx 1.92 points and GPTZero 20.50 points. Turnitin 0% (n = 220): Academi.cx 0.18 points and GPTZero 9.30 points. Turnitin 20–99% (n = 110): Academi.cx 3.95 points and GPTZero 36.32 points. Turnitin 100% (n = 99): Academi.cx 3.53 points and GPTZero 27.80 points. Blue represents Academi.cx and gray GPTZero. Lower is better. All bars share a 0 to 40 point scale.

On the 0% documents, Academi.cx averaged 0.18 points of error, GPTZero averaged 9.3. The bigger issue shows up on the other end, on the 100% documents Academi.cx averaged 3.53 points off versus GPTZero at 27.8, and in the mixed bucket GPTZero's error climbed even higher to 36.32 against Academi.cx's 3.95. GPTZero's error grows the most exactly where the AI content is heaviest, the opposite of what you'd want from a detector.

False flags and missed AI

Academi.cx had a 0% false flag rate and a 0% missed AI rate across all 429 documents, meaning it never scored a genuinely 0% document above 20%, and never scored a genuinely 100% document below 80%. GPTZero came in at a 13.2% false flag rate and a 33.3% missed AI rate, the highest missed AI rate of any tool covered in this benchmark so far, meaning one in three fully AI written documents scored below 80% AI on GPTZero.

Sentence level accuracy

We also checked how often each tool correctly classified individual sentences as AI written or human written, matched against Turnitin's own sentence level calls. Across all 429 documents, Academi.cx correctly classified 96.64% of sentences. GPTZero came in at 72.67%.

The takeaway

Across all 429 documents, Academi.cx stayed close to the actual Turnitin percentage on every measure checked, while GPTZero's biggest weakness showed up on the documents that matter most for catching AI use, the fully AI written and mixed content, where its error and missed AI rate both ran highest in this benchmark.

See the full benchmark results.

Related reading