ISTANBUL

Google’s upcoming large language model (LLM) for artificial intelligence (AI), Gemini 3, reignited the AI race with its surpassing performance in independent tests, surpassing ChatGPT.

Google is launching its new generative AI model. Gemini, currently boasting 650 million monthly users, will be upgraded to handle tasks such as coding, image recognition, and content creation with fewer prompts than before.

Gemini 3 achieved the highest score ever in the Humanity’s Last Exam analysis, with a grade of 37.5%, while ChatGPT 5.1 scored 26.5%.

Gemini 3’s Deep Think mode takes the model’s logic to the next level, as well as its grade to 41% in the Humanity’s Last Exam, meanwhile, ChatGPT-5 Pro scored 30.7%.

Google’s new model excelled in deep thinking, achieving 93.8% on the GPQA Diamond test and 45.1% on the ARC-AGI-2 test.

Gemini’s performance, however, still did not surpass ChatGPT in visibility and market share.

ChatGPT has had a significant advantage over other AI tools in monthly visitor traffic, according to data from the Semrush platform.

Google’s search engine dominated web traffic with 101 billion visits, followed by YouTube with 47 billion, Facebook 10 billion, Instagram 5.7 billion, and ChatGPT with 5.2 billion.

ChatGPT also accounted for 81.3% of worldwide web AI traffic, followed by Perplexity with 11.1%, Microsoft Copilot with 3.4%, and Google Gemini with 3%, according to Statcounter.

Türkiye’s AI web traffic also concentrated on ChatGPT with a 91% share, followed by Google Gemini with 5%.

Generative AI is still expected to grow, as a report by McKinsey & Company in June 2023 showed that generative AI is expected to add $2.6-4.4 trillion to the global economy annually.

Over 80% of businesses will start using generative AI models by 2026, another report by Gartner showed. This rate was below 5% at the beginning of 2023.