Gemini 3 surpasses ChatGPT and redefines leadership in artificial intelligence

The recent launch of Gemini 3, Google's most advanced language model, has redefined the global competition in artificial intelligence and placed the company in a leadership position it hasn't held for years. According to The Wall Street Journal, the new version of Gemini outperformed ChatGPT and other rivals in recognized industry benchmarks, solidifying its status as the most powerful AI chatbot to date.

The magnitude of this advance transcends the technological aspect. According to CNBC, Google's renewed position in artificial intelligence has generated concern among its main competitors, especially OpenAI. An internal memo cited by The Information and reported by CNBC reveals that Sam Altman, CEO of OpenAI, warned his team about the "complicated environment for a while," anticipating temporary economic challenges stemming from Google's progress.

A comparison between Google and OpenAI after the launch of Gemini 3: Google's most advanced AI model consolidates its financial strength and user base, while OpenAI faces economic challenges and increasing competitive pressure.

Altman warned that these difficulties could have existential consequences for OpenAI, which is not yet profitable and depends on constant funding rounds, unlike Google, which sustains its AI strategy thanks to the profits from its search engine.

Google's financial performance reinforces its strength in the sector. According to CNBC, the company posted a gross margin of 32%, cash reserves of $112 billion, and free cash flow of $24 billion in the last quarter.

These figures contrast sharply with OpenAI's situation, which is facing a rapid cash burn and the need to continuously raise funds to remain competitive.

Analysts cited by CNBC believe Google can withstand a slowdown in the sector, while a potential correction in the AI market could pose significant risks for other players.

Gemini 3's success is demonstrated by its performance in specialized tests. According to The Wall Street Journal, the model outperformed others in more than a dozen assessments measuring technical knowledge, logic puzzle solving, mathematical problems, and image recognition. Only in one coding test did Anthropic's Claude Sonnet 4.5 model score higher.

Tulsee Doshi, Gemini's senior director of product management, highlighted to The Wall Street Journal the model's ability to operate in languages underrepresented on the internet, such as Gujarati, and its outstanding performance in the Vending Benchmark test, used to evaluate planning and tool usage over time.

“Vending Bench (a simulated benchmarking environment designed to test the capabilities of artificial intelligence agents) reflects one of the aspects we wanted to transform and enhance with this model, which is the improved use of tools and planning,” Doshi explained to The Wall Street Journal.

The launch of Gemini 3 boosted the use of related products, such as Nano Banana, an image-generating tool that contributed to Gemini's rapid user growth this year. According to The Wall Street Journal, Gemini's monthly user base reached 650 million, up from 450 million in July, although ChatGPT remains the most popular chatbot, with 800 million weekly users.

However, Gemini 3's advancements could make it the preferred choice for various tasks, according to users and analysts consulted by The Wall Street Journal.

Google's strategy included a deep internal restructuring. CEO Sundar Pichai and other senior executives reorganized teams, eliminated silos, and centralized model development, according to employees cited by The Wall Street Journal. Furthermore, co-founder Sergey Brin resumed an active role overseeing AI efforts.

During its annual developers conference in May, Google unveiled a new generation of artificial intelligence products and a revamped version of its search engine with AI Mode, which responds in a conversational format. This advancement restored confidence among some investors, although the stock price remained unchanged throughout the summer. Analyst Michael Nathanson told The Wall Street Journal, “Wall Street was debating whether they were going to fall victim to AI.”

The development of Gemini 3 focused on improving its analysis and content generation across multiple formats—text, images, audio, video, and code—and on strengthening its reasoning capabilities to position itself as an effective personal assistant, particularly in programming. Following the launch, Gemini 3's results from 20 benchmark tests were released, in which the model outperformed ChatGPT and Anthropic in most categories, according to The Wall Street Journal.

The integration of Gemini 3 into Google's search engine via AI Mode marked the first time a new model was incorporated from day one. Robby Stein, vice president of product for search, told The Wall Street Journal how he used AI Mode to explain the lift force of an airplane to his seven-year-old son and received an interactive simulation that allowed him to modify the airflow and raise the wing. "I was surprised to see that it can really present the information in the best way depending on the question," Stein told The Wall Street Journal.

Google's advantage lies not only in the quality of its model but also in its ability to distribute it globally. CNBC highlights that the company has the infrastructure, ecosystem, chips, models, and products to bring artificial intelligence to billions of users, even those not actively searching for a chatbot.

Global distribution and massive integration strengthen Google's position against competitors who lack such a broad and established user base.

The shift in dynamics is evident in the industry. Satya Nadella, Microsoft's CEO, said he wanted to "make Google dance," as CNBC recalled. Today, in Mountain View, the music is still playing for Google.