Chatgpt attracts more traffic than the next nine AI tools together, with 5.5 billion visits crushing Gemini and Claude.
The Chinese startup Deepseek exploded from 33.7 million to 436 million monthly visits in just four months, operating on 1/30 the costs of Western models.
Traditional sectors are bleeding users: freelance platforms fall 14%, educational sites with 19%and Chegg collapsed 64%while students turn to AI.
The latest traffic data from SimilseWeb reveals an uncomfortable truth about the generative AI market: although the world debates that AI model is technically superior, Chatgpt has already won the user adopt war with a margin so enormously, tart conventional competition statistics.
With 5.5 billion visits in May 2025, Chatgpt orders around 80% of all generative AI traffic. To understand this scale: Chatgpt gets more visits than Google’s Gemini, Deepseek, Grok, Perplexity and Claude together. Then doubled. Then add a few more millions for a good measure.
Image: Similar web
Chatgpt rose at the end of March after 500 million weekly active users and the Chatgpt mobile app was on average more than 250 million monthly active users between September and November 2024. The partnership of OpenAI with Microsoft certainly helped, but the scale suggests slightly more fundamental: Chatgpt has become the standard AI -assistant millions of millions of millions of millions of millions of millions.
What is especially interesting is the resilience of Chatgpt. Despite a short dip in traffic at the beginning of 2025, OpenAi quickly turned around and rose back to new heights.
China’s Deepseek: neither disappeared nor forgotten
Another surprising data point: the Chinese Startup Deepseek-Bans on most American devices issued by the government and many institutions fear that the nutritional information to China was wax in January of 33.7 million monthly users to 436 million visits by May. That is a 13x increase that would cause venture capitalists to cry.
Although OpenAi has collected billions of dollars to collect as many GPUs as possible to feed its models, Deepseek works against a fraction of the costs of Western models; The inpoertokens cost $ 0.55 versus OpenAIs $ 15. The spicy upStart has achieved this while it was forced to use lesser Nvidia H800 chips due to export restrictions, the computational equivalent of winning a Formula 1 race with a Toyota Corolla.
Image: Artificialanalysis
The deep dominance of Deepseek tells the story: Deepseek builds up a rich in markets that have hardly touched Western AI companies. The top three markets, according to Similseb – China, India and Indonesia – make more than 51% of its users.
The best of the rest: Google and Anthropic
For a company of $ 2 trillion that processes 3.5 billion searches every day, the AI performance of Google is quite pedestrians. The 527.7 million visits of Gemini are hardly the depth of the deep road, despite any conceivable advantage: billions in financing, integration into the Ecosystem of Google and access to more users than any platform on earth – a monopolization of the AI browser and partnerships with important brands to increase the acceptance of the model.
However, it should not be postponed: Gemini registered 284 million visits in February, which means that the model is growing in popularity – a good metric when it becomes a trend, given the current state.
Perhaps the most unexpected finding is the poor performance of Claude. Despite the support of Amazon and Google, technical superiority in many benchmarks and constant praise from AI researchers and people in the academic world, Claude attracted slightly less than 100 million visits in May 2025.
From the beginning of 2025, Claude had 18.9 million monthly active users worldwide, and although the Golf is certainly a large increase in traffic, these figures were turned out to be compared to the billions of Chatgpt.
The disconnection between the technical possibilities of Claude and the acceptance of users is interesting. Claude 4 Opus scored better than his competition with complex reasoning and even creative tasks, but it has not succeeded in translating this into massive adoption, which shows that the best product is usually not the one that everyone loves.
Although Meta AI exists, the current use as a primary AI tool can be considerably lower compared to chatgpt. Many users are currently not attracted by the chatbot from Zuck as their preferred choice for AI-related tasks.
The important contribution from Meta to the Open-Source community, in particular through the LAMA models, is remarkable. However, the nature of open-source use is distinguished from direct user interaction.
The AI-driven massacre
While AI companies are fighting for position, traditional internet companies see their rich crumbling. Similar web, for example, reports that the traffic of Chegg collapsed 64%, because students discovered that Chatgpt gives free homework help, Quora fell 51%and freelance platforms such as Fiverr fell 14%.
On the other hand, Devops and Code Completion tools with 41%on an annual basis, speech generation platforms grew by 14%and the automation tools rose by 12%.
The similar web data provides a brutal truth to the AI industry: technical superiority means little without acceptance of users. Claude may impress researchers, but Chatgpt impresses everyone. Gemini may have the distribution of Google, but distribution without differentiation is not interesting.
Chatgpt won by being the first, being good enough, and above all, what users actually wanted – a conversation -ai that just works. While competitors concentrated on benchmark scores and safety protocols, chatgpt was focused on being useful. The visits of 5.5 billion say that the strategy worked.
So OpenAi can lose in some benchmarks, but in the race for global adoption it has already exceeded the finish line. The real competition is now for second place.
Published by Andrew Hayward
Generally intelligent Newsletter
A weekly AI trip told by Gen, a generative AI model.
