Which AI is the smartest?

Who’s the Smartest AI in 2026?

As we move deeper into 2026, the race for AI supremacy has reached a new level of intensity. According to the latest TrackingAI Mensa Norway benchmark released in April, two models are now neck-and-neck at the top: Grok-4.20 Expert Mode and OpenAI’s GPT 5.4 Pro (Vision). Both have achieved an impressive score of 145—a mark once thought years away.

This tie is more than just a headline; it signals a turning point. For the first time, the leading AI models are so close in performance that small refinements could shift the rankings overnight. The gap between first and fifth place now spans fewer points than seen in previous years, reflecting how rapidly the field is advancing.

What sets these top models apart isn’t just raw reasoning power, but their ability to interpret complex queries, adapt to context, and handle multimodal inputs with human-like fluency. GPT 5.4 Pro (Vision) excels in visual reasoning and real-time data synthesis, building on OpenAI’s long-standing lead in natural language understanding. Meanwhile, Grok-4.20—developed by xAI—has gained ground through aggressive optimization and deep integration with real-time knowledge sources, giving it an edge in up-to-the-minute awareness and analytical depth.

Still, intelligence benchmarks like Mensa Norway’s are just one lens. They measure cognitive ability under test conditions, but real-world performance depends on reliability, safety, and usability—areas where differences still matter greatly.

As the leaderboard tightens, the bigger story isn’t who’s #1, but how quickly the entire field is rising. With models now operating at levels comparable to human top percentiles, the focus is shifting from pure IQ to practical wisdom—how AI makes decisions, handles uncertainty, and aligns with human values.