Eastleigh Voice attributed the ranking to AITools.xyz and said its estimates were modeled from SEMrush and Ahrefs data. The same June figures, down to Microsoft Copilot’s 92.72 million visits, appear in OneLittleWeb’s global traffic ranking, which tracks more than 9,500 AI tools. That matching table confirms the broad result: ChatGPT is far ahead in direct web traffic, Gemini is second at 1.14 billion visits, Claude is third at 967.91 million, and Copilot ranks 14th in the broader AI-tools list.
The more useful finding for Windows and Microsoft 365 users is the limit of the ranking. Copilot’s position is a website-traffic result, not a measure of Copilot use across Microsoft’s products. Treating it as a full market-share table would undercount the services most deeply integrated into operating systems, office software and enterprise workflows—and would overstate the value of a click to a public chatbot page.
ChatGPT’s lead is real in this particular measurement
The arithmetic behind the ChatGPT headline checks out. The top 15 entries together total roughly 10.02 billion June visits, putting ChatGPT’s 5.32 billion at about 53% of that selected group. The remaining 14 entries total approximately 4.70 billion.
Gemini’s 1.14 billion visits and Claude’s 967.91 million are large numbers, but neither comes close to the ChatGPT site’s web footprint in this dataset. The rest of the ranking fragments quickly: Canva recorded 759.76 million visits, Google Translate 343.02 million, DeepSeek 318.81 million and Grok 189.89 million.
That is meaningful evidence that ChatGPT remains the default destination for people who deliberately open a browser and seek a general-purpose AI assistant. It is also consistent with Pew Research Center’s February 2026 survey of U.S. adults: 44% said they used ChatGPT, compared with 24% for Gemini and 17% for Copilot. Pew’s survey and the traffic rankings measure different things, but both point to a large consumer-awareness and usage advantage for OpenAI’s product.
The ranking also demonstrates how much of the AI market is still centered on a handful of well-known sites. Character-focused services JanitorAI and Character AI received 173.17 million and 165.46 million visits, respectively. Their combined traffic exceeded the combined June estimates for Perplexity and Microsoft Copilot in the same table.
That result says more about the web’s consumer behavior than about enterprise deployment. A person can open a character-chat site dozens of times in a month; a Microsoft 365 user may encounter Copilot during document drafting, meeting summaries, Outlook triage or a Windows task without ever visiting a standalone Copilot site.
The list is broader than “AI chatbots”
The top-15 ranking is often described as a list of AI tools, but its categories are inconsistent enough to make head-to-head product conclusions risky. Google Translate ranks fifth, Canva ranks fourth, Zoom ranks 13th and Notion ranks 12th. All have AI features, but they are established products with extensive non-AI use cases.
Google Translate’s 343.02 million visits, for example, cannot be cleanly read as demand for generative AI in the way a visit to ChatGPT, Claude or Copilot might be. The same issue applies to Canva and Zoom: users may arrive for conventional design, videoconferencing or collaboration work rather than a discrete AI interaction.
OneLittleWeb’s own classification reflects that mixture. It labels Copilot as both a text-generation and chatbot product, but includes Zoom as an AI note-taking tool and Notion as a productivity or project-management service. The methodology can support a directory-style view of web destinations that offer AI. It cannot establish a clean ranking of competing assistants, let alone a ranking of model capability or business adoption.
For a more focused chatbot comparison, OneLittleWeb’s June table puts ChatGPT first, Gemini second, Claude third, DeepSeek fourth, Grok fifth, Perplexity sixth and Microsoft Copilot seventh. Copilot’s traffic estimate remains 92.72 million, but the narrower category removes Canva, Google Translate, Zoom and other products whose traffic is not principally driven by chatbot use.
The crucial caveat is that even this cleaner comparison still measures site visits. It does not count a user invoking Copilot from Word, Excel, Teams, Windows, Edge, Visual Studio or another Microsoft product surface. Nor does it capture API calls, which are particularly relevant for developers and companies deploying AI inside internal tools.
Copilot’s web number should not be read as its installed-base reach
Microsoft Copilot’s 92.72 million June website visits place it far below ChatGPT’s 5.32 billion in the shared estimate. The gap is too large to dismiss as statistical noise, and it shows that Microsoft has not built a standalone consumer web destination on the scale of ChatGPT.
Yet the number carries a narrower implication than the headline suggests. Microsoft’s strategy has been to put Copilot inside products customers already use rather than to make a browser tab the only—or even primary—place to engage with it. That creates a measurement problem for any ranking based solely on public web domains.
An employee who uses Copilot in a Microsoft 365 app might generate no visit to the public Copilot website. A Windows user who asks Copilot for help inside the operating system may not show up in the same count either. Conversely, a ChatGPT user who repeatedly returns to its website will register multiple visits even if they are one person using the service for a few minutes at a time.
This does not mean Copilot is secretly as large as ChatGPT. There is no public number in the submitted ranking that can support that conclusion. It means the ranking cannot answer the question.
For IT administrators, the distinction is practical. Website popularity can indicate where employees are likely to bring their own AI habits into work, which matters for data-governance, browser controls, identity policy and approved-tool decisions. It does not reveal whether a company’s licensed Microsoft 365 Copilot deployment is active, useful or delivering value. Those questions require tenant-level usage reports, adoption telemetry, support trends and task-specific outcomes.
Different rankings produce materially different totals
The strongest reason not to treat 5.32 billion visits as a definitive census is that other June 2026 web-traffic rankings report substantially different totals while preserving the same broad ordering.
AICPB’s June global AI ranking estimated ChatGPT at 5.56 billion website visits, Gemini at 2.95 billion, Claude at 965.59 million and Microsoft Copilot at 121.12 million. Its chatbot-only listing reported the same ChatGPT, Gemini, Claude and Copilot figures. Those totals are not minor rounding differences: Gemini’s estimate is more than twice the 1.14 billion number in the AITools.xyz/OneLittleWeb-derived table, while Copilot’s is roughly 31% higher.
The disagreement does not disprove either ranking. Traffic-estimation firms use different panels, domains, categorization rules, bot filtering, country coverage and methods for combining subdomains or related properties. AICPB also includes products absent from the top-15 AI-tools list, including New Bing and GitHub Copilot in its broader global ranking.
It does show that the popular “more than the next 14 combined” framing is specific to one selected list and one traffic model. It should not be repeated as a universal measure of the entire AI market.
The same applies to the claim that 12 of the 15 most visited AI tools are U.S.-based. It is accurate for the submitted list, where Canva, DeepSeek and Freepik’s Magnific are the three non-U.S. entries. But it reflects which tools were included and how their traffic was assigned—not a complete accounting of AI use worldwide. A chatbot-only ranking that includes Chinese services such as Doubao, Qwen, Kimi and Tencent Yuanbao produces a visibly different geographic picture.
What the figures mean for Windows users and IT teams
The June traffic data supports a straightforward conclusion: ChatGPT remains the dominant public web destination for AI, and Microsoft Copilot’s standalone website draws a much smaller audience. It does not establish that ChatGPT has the same lead inside corporate work, on managed Windows devices, or in the productivity software where Microsoft is embedding Copilot.
For organizations, the immediate risk is assuming that an approved Copilot deployment eliminates employee use of ChatGPT, Gemini or Claude. The public-web figures suggest the opposite: users are already familiar with competing services and will continue to reach them through browsers unless policy, training and a useful sanctioned alternative make the approved route practical.
For Microsoft, the 92.72 million estimate is a reminder that Windows and Microsoft 365 distribution do not automatically turn into destination-site traffic. That may be intentional. But it also means web rankings will continue to favor products whose primary interface is a public browser tab.
ChatGPT’s 5.32 billion-visit headline is therefore a real signal of consumer web demand, not a final scoreboard for AI adoption. The most consequential usage is increasingly the kind that a public traffic ranking cannot see.