Comparing the IQ of AI Models

What We’re Showing

This infographic ranks 24 leading AI models by their performance on the Mensa Norway Intelligence Quotient (IQ) Test, a high-difficulty cognitive benchmark used to evaluate IQ. For context, the average human IQ score ranges from 90 to 110, while a score above 130 is typically considered genius level.

The data is sourced from Tracking AI.

Key Takeaways

• OpenAI’s o3 model leads all contenders with an IQ of 135, placing it in the genius range.
• Anthropic's Claude-4 Sonnet (127) and Google's Gemini 2.0 Flash (126) also significantly outperform average human intelligence, along with Gemini 2.5, OpenAI o4 mini, and Claude-4 Opus.
• The top 10 are text-only models, while multimodal and vision models rank lower in terms of IQ.
• Vision-first models such as GPT-4o (Vision) and Grok-3 Think (Vision) scored just 63 and 60, respectively—the lowest of all tested.