Saad Ullah
⬤ Google's Gemini 3 Pro has drawn attention after it reached IQ scores that place it ahead of rival systems in reasoning tests. The model achieved about 130 in an offline run and 142 on the Mensa Norway benchmark. Those values sit in the uppermost band of human performance when judged against classical IQ scales.
⬤ A result close to 130 ranks in the top 2 % of human candidates, while 142 falls within the top 0.3 %. For reference the mean human IQ equals 100 and most PhD holders score between 120 and 130. The comparison table displays Gemini 3 Pro at the front of a large set of AI models, ahead of Grok-4 Expert Mode, Claude-4.1 Opus besides OpenAI GPT-5 Thinking.
⬤ The direct comparison shows that the majority of frontier AI models group between 110 and 130, with only a handful rising above 135. Gemini 3 Pro's 142 on the Mensa Norway scale remains the highest score recorded for any system tested. The narrow spread demonstrates the speed at which the field is advancing toward deeper reasoning capacity.
⬤ The benchmark outcomes are altering views on AI capability. Although IQ tests were created for people, they now serve as a frequent measure for contrasting reasoning strength across models. Gemini 3 Pro's margin underscores Google's advances in model design and sets a fresh bar for the next stage of AI development.
Saad Ullah
