Research
Data we measure ourselves, published with the method, the raw data and a suggested citation. Free to reuse under CC BY 4.0.
Study · 2026-10-21
AI vs human IQ 2026: how AI models score on a culture-fair IQ test
5 AI models took the same 25-puzzle test people take, 5 times each, scored with the same Rasch scorer. Mean estimated IQs ranged from 124 to 135.
How we publish research
- Every study states its method in enough detail to repeat it, including prompts and settings where they apply.
- The data behind every table is downloadable.
- Limitations are listed on the study page, next to the results, not in a footnote.
- Reruns are dated, and corrections are logged on the corrections page.
The scoring behind every study is described on the methodology page. For press questions, see the press kit or email support@iqoctopus.com.
Related
- No. 01AI vs human IQ 2026Same puzzles, same scorer as people. Methods, data and limits.
- No. 02AI leaderboardThe same puzzles and the same scoring as people. Measured, not guessed.
- No. 03MethodologyItem templates, Rasch EAP scoring, the confidence range and the limits.
- No. 04PressBoilerplate, key facts, story angles, logos and a contact.