AI Maxes Out Mensa Norway IQ Test at 151, Beyond Human Ceiling
AI has maxed out the Mensa Norway IQ test, formally breaking the ceiling of this human assessment. Per the TrackingAI leaderboard, Claude Fable 5.1 and vision-enabled GPT-6 Astra answered all 35 questions of the 25-minute timed test correctly, converting to a score of 151—above the human score-reporting cap of 145—amid a seven-run upward streak. Musk reshared the news.
Confirmed
- Curves compiled by AISafetyMemes and others show the best AI scoring about 64 (near random guessing, comparable to cognitively impaired humans) three years ago in 2023, reaching the average human two years ago, genius level a year ago, and now the maximum 151.
- 新智元 citing TrackingAI: Claude Fable 5.1 and GPT-6 Astra scored 151, exceeding the human ceiling of 145.
Why it matters
- Within three years, AI moved from far below human level to beyond the scale on the same standardized test—a vivid illustration of rapid gains in abstract reasoning, cited by sharers like @SydSteyerhart and @CurieuxExplorer as evidence of accelerating AI capability.
- Caveat: a single online IQ test is not a comprehensive measure of general intelligence; the test being "broken" mainly means its difficulty no longer discriminates among top models.
Timeline
- 2023: best models around 64 (cognitively impaired level)
- 2024: average human level
- 2025: genius level
- Sept 2025: maximum 151, "off the chart"
2026-09-25 ~ 2026-09-26 · 5 related posts
Primary sources
- AIs hit the maximum 151 IQ on Mensa Norway test as Musk flags the trend — elonmusk · 2026-09-25
- AIs hit the ceiling on Mensa Norway IQ test: max score 151, up from 64 three years ago — CurieuxExplorer · 2026-09-25
- AIs now max out the Mensa Norway IQ test at 151 — from 64 three years ago — SydSteyerhart · 2026-09-25
- [source] Claude and GPT Hit Mensa Test Ceiling: Perfect Scores for 151 IQ, Beyond Human Scale — 新智元 · 2026-09-26
1 near-duplicate retellings: coherence