Google's Gemini 3 Sets New AI IQ Milestone at 130, Raising Bar for AI Intelligence

A New Benchmark for AI Intelligence

Google’s Gemini 3 Pro has achieved an IQ score of 130, marking a significant leap in artificial intelligence capabilities. This milestone, revealed by Tracking AI, positions Gemini 3 at the top of the AI intelligence rankings, surpassing competitors like OpenAI’s GPT 5.2 Pro and xAI’s Grok 4 Expert Mode. The rapid advancement in AI IQ scores highlights both the pace of innovation and the growing potential of AI systems to approach human-level reasoning.

In just two years, AI IQ scores have seen a dramatic increase, with top models gaining roughly 30 points since 2024. This progress is not only impressive but also measurable, as AI systems now demonstrate problem-solving abilities comparable to human intelligence. The steady climb in scores suggests a consistent trajectory toward even greater AI capabilities, with experts predicting that AI could reach top human performance on visual IQ tests by the end of 2027.

The Path to AI Intelligence Growth

The journey to this milestone has been marked by steady and linear progress. In February 2024, AI IQ scores were around 60, barely distinguishable from random guessing. By October 2025, the scores had surged to 105, nearly matching the average human IQ of 104. This growth was not explosive but rather a consistent upward trend, with top scores increasing by an average of 2.5 points per month.

Tracking AI’s methodology plays a crucial role in validating these scores. By using offline IQ tests designed by Mensa members and ensuring the questions are not part of any known training datasets, the platform provides a reliable measure of AI reasoning abilities. This approach minimizes the risk of AI systems gaming the tests, offering a clearer picture of their true capabilities.

Implications for the Future of AI

The rise in AI IQ scores has sparked discussions about the future of AI and its potential impact on society. While the scores indicate rapid progress, experts like Roth emphasize that significant barriers remain before AI can take on fully autonomous, human-like roles. Instead, the current phase is seen as a window of preparation, allowing society to adapt to the evolving landscape of AI intelligence.

John Warner, a senior research fellow at MIT, has described AI IQ as a potential new standard for measuring AI progress, akin to Moore’s Law for computing. This perspective underscores the importance of tracking AI intelligence not just for its novelty but for its correlation with real-world problem-solving abilities. As AI continues to advance, the focus will likely shift toward harnessing this intelligence responsibly and effectively.