Key Findings on AI Mensa Norway Scores (2025–2026)

• Top Performers (2026): According to TrackingAI's April 2026 benchmark, models like OpenAI GPT-5.4 Pro and Grok-4.20 Expert Mode have hit scores of 145–150 on the test.

• Rapid Growth: AI performance on this test has increased exponentially. In early 2025, models scored around 117; by April 2025, OpenAI's o3 reached 136, exceeding the top 2% threshold for human Mensa entry.

• Offline/Contamination Check: While some argue AI excels due to training data contamination, top models still scored 116 on an "offline" version of the test, confirming high-level reasoning capabilities.

• Pattern-Recognition Focus: The test is highly visual, relying on 35 progressive matrix patterns. Because AI is trained specifically for pattern recognition, these tests are highly optimized for them.

Performance Comparison (Approximate)

• Average Human: IQ 100

• Mensa Norway Entry Threshold: Approx. IQ 132–135

• GPT-4 (2024): ~85–96 IQ

• OpenAI o3 (April 2025): 136 IQ

• GPT-5.4 Pro/Grok-4.20 (April 2026): 145–150 IQ

While these scores actually demonstrate incredible pattern recognition in the AI models tested so far, they do not necessarily equate to human-like general intelligence, as LLMs indeed can often solve visual puzzles better through specialized training.

Then, ¿are you just average or are you smarter than your favorite AI model?

««If you appreciate what I've said, don't upvote me, be smarter!»»

^{Leave a comment. Share your experiences and feedback. ¡Be part of the conversation!}

«««-$-»»»

"Follows, Comments, Rehives & Upvotes will be highly appreciated"

Cheers!