For more than three years, asking a question to an AI chatbot meant, in practice, asking ChatGPT. That changed sometime around May 2026. According to Sensor Tower’s State of AI 2026 report, OpenAI’s chatbot closed out May with 46.4% of the global AI-assistant app market, the first time its share has dropped below half since the product launched in November 2022. Google’s Gemini climbed to 27.7%. Anthropic’s Claude reached 10.3%. TechCrunch, which first reported the Sensor Tower figures on June 16, 2026, still called ChatGPT “the most popular assistant globally,” and by raw user count that holds up: OpenAI says ChatGPT crossed 900 million weekly active users on February 27, 2026, and became the fastest app in history to reach 1 billion monthly users by June 2026. The picture got murkier still by August 2026: First Page Sage’s own tracking put ChatGPT’s AI-search market share back up at 52.6% that month, a reminder that whether the majority has actually been lost still depends on exactly which methodology is doing the counting. Market share and user count are telling two different stories right now, and the gap between them is the real story. Here is what the numbers show, why different trackers disagree so sharply about the size of the drop, and what it means for a market that spent three years looking like a one-company race.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
ChatGPT’s Market Share Falls Below 50% for the First Time Since Launch
The headline number comes from Sensor Tower, the mobile and app analytics firm behind an annual State of AI report. Its 2026 edition, covering data through the end of May, put ChatGPT’s share of the global AI-assistant app market at 46.4%. Gemini followed at 27.7%, Claude at 10.3%, and the remaining assistants Grok, Perplexity, DeepSeek, and Meta AI were all under 5%, according to TechCrunch’s June 16, 2026 report.
That 46.4% figure matters less as a standalone stat than as the end of a streak. ChatGPT had held more than half the app market continuously since its November 2022 launch and stayed above 50% as recently as January 2026, when Statcounter Global Stats put its worldwide AI-chatbot share as high as 80.49%, per The Digital Elevator’s tracking of that data. Losing that majority, even to a plurality still nearly twice the size of its nearest rival, is a symbolic marker. The period when “AI chatbot” and “ChatGPT” were functionally synonymous in app-usage terms is over.
TechCrunch’s framing of the shift was direct: users are migrating between assistants like Gemini, Claude and xAI’s Grok. That migration didn’t happen overnight. Fortune, citing mobile intelligence firm Apptopia, had already flagged the trend back in February 2026. The slide continued from there: ChatGPT’s global app-market share is reported to have fallen from 65.3% in December 2024 to 52.8% by December 2025, before dropping to Sensor Tower’s 46.4% by the end of May 2026. Two different analytics firms, using two different panels, reached broadly the same conclusion months apart: OpenAI’s app-market dominance was eroding well before May, and it kept eroding through the spring. For ongoing coverage of how the underlying models stack up, see our AI models hub.
App Share vs. Web Traffic: Why Two Trackers Tell Different Stories
Read a different report and ChatGPT never came close to losing its majority. StatCounter’s Global Stats service, which tracks web-traffic referrals across millions of sites, put ChatGPT at 76.87% of worldwide AI-chatbot web visits in June 2026, with Gemini at 7.94%, Perplexity at 7.91%, Claude at 3.74%, Microsoft Copilot at 3.49% and DeepSeek at just 0.03%. StatCounter’s own tracker then ticked back up to 77.92% in July 2026, a reminder that even within a single web-traffic methodology the month-over-month trend isn’t a clean straight line down. That’s not a rounding difference from Sensor Tower’s numbers. It’s a different picture of the market entirely, and both readings are defensible, because they are measuring different things.
App Installs and Sessions, Not Web Visits
Sensor Tower’s 46.4% figure tracks usage of native mobile and desktop apps, pulled from app-store intelligence and device panels. It captures how people use AI assistants that live on a phone’s home screen, which is where a large share of Gemini’s growth is coming from, given its default placement on Android devices.
Browser Visits Skew Toward ChatGPT’s Core Audience
StatCounter and Similarweb, by contrast, track visits to web domains like chatgpt.com, gemini.google.com and claude.ai. That method tends to favor whichever product has the stickiest desktop and browser-based habit, historically ChatGPT among developers, writers and researchers doing sustained work at a keyboard. Similarweb data cited by Yahoo Finance had already put ChatGPT’s global generative-AI web share at 53.7% in April 2026, and a separate Similarweb-based read from Momentic Marketing put ChatGPT’s worldwide web-visit share at 53.9% the following month, down from 79.0% a year earlier, a steep decline but nowhere near Sensor Tower’s sub-50% app number. The exact percentage changes depending on who is counting and what they are counting. The direction does not.
Gemini’s Surge to 27.7%: How Google Converted Distribution Into Adoption
Whichever dataset gets used, Gemini comes out as the biggest winner. Sensor Tower has it at 27.7% of the app market at the end of May, up from a low-teens share a year earlier. Apptopia’s numbers, cited by Fortune, showed Gemini’s app share climbing from 14.7% in January 2025 to 25.2% in 2026. A separate Similarweb-based tracker published by SudoFlare put Gemini’s web-visit share at 27.4% in June 2026.
The consistent driver across every source is distribution rather than a single breakthrough model. Gemini ships as the default assistant on Android phones, inside Chrome, and across Google Workspace, putting it in front of billions of people who never had to seek it out. That’s a structural advantage neither OpenAI nor Anthropic can match without an operating system or hardware platform of their own. For a look at how the underlying models compare on cost and benchmark scores rather than distribution, see our breakdown of Sonnet 5 vs GPT-5.6 vs Gemini 3.1 Pro.
Claude’s Enterprise-Led Breakout: From 10% Share to Serious Contender
Anthropic’s growth is smaller in absolute share but faster in percentage terms. Sensor Tower puts Claude at 10.3% of the app market. SudoFlare’s numbers are more dramatic: Claude’s monthly web visits reportedly grew 306% in a single quarter, from 203 million to 824 million, pushing its worldwide web-visit share to roughly 8.2% by that count.
The more telling number sits outside consumer market share entirely. Data cited by EnterpriseDNA’s analysis of the fracturing chatbot market found that Claude wins roughly 70% of new enterprise deals when businesses choose directly between Anthropic and OpenAI. That’s a head-to-head procurement number, not a market-share estimate, and it points to where Anthropic has focused: coding tools, agentic workflows and enterprise contracts rather than mass consumer reach. That strategy shows up more clearly in Anthropic’s product releases than in any single traffic chart. See our coverage of Claude Sonnet 5’s debut for how that model line has positioned itself on price and benchmark score.
Grok, DeepSeek and Perplexity: The Rest of the Field
Below the top three, the numbers get harder to pin down. TechCrunch’s Sensor Tower summary lumps Grok, Perplexity, DeepSeek and Meta AI together as each holding under 5% of the app market, without individual breakouts. StatCounter’s web-traffic numbers are more specific but tell an odd story: Perplexity at 7.91% (ahead of Claude’s 3.74% on that measure), Microsoft Copilot at 3.49%, and DeepSeek at just 0.03%. Momentic’s separate read puts DeepSeek at 4.1% and Grok at 2.8% of worldwide web visits, more than a hundred times StatCounter’s DeepSeek figure for the same rough window.
That kind of spread, for the same company, in the same rough time period, is a useful reminder of how noisy this category of data still is. Regional traffic patterns, bot-filtering rules and which properties count as “AI chatbot” traffic all shift the outcome. No single number for DeepSeek or Grok in this piece should be read as precise. What every source agrees on is that the market went from three real options to six or seven, a genuine change from how the field looked in 2024. DeepSeek’s cost profile is still competitive on paper, a point we cover in our best AI model for coding comparison.
Enterprise vs Consumer — and Why the US Market Looks Different
Zoom into the United States and the picture shifts again. Momentic’s May 2026 breakout shows ChatGPT holding 58.3% of US web-visit share, well ahead of its global web-traffic numbers from the same firm, with Gemini at 19.3% and Claude at 13.4%. On the app side the domestic picture looks tighter: Apptopia data cited by Big Technology put ChatGPT’s U.S. mobile-app share at just 45.3% in January 2026, already below the halfway mark months before Sensor Tower’s global app figure crossed that same threshold. OpenAI’s home market is holding up better on web traffic than on mobile, or than its international footprint overall, likely reflecting brand loyalty built up over three-plus years as the default choice for English-language users.
Enterprise tells the opposite story. The consumer numbers above are dominated by ChatGPT’s large free-tier user base, people asking one-off questions in a browser or app. When a business is choosing a primary AI vendor for internal tools, coding assistants or customer-facing products, Claude reportedly wins the deal about 70% of the time it goes head-to-head with OpenAI. That split, OpenAI strong in US consumer usage, Anthropic stronger in enterprise procurement, means the “market share” headline number depends heavily on which market is being described. A consumer comparing subscription plans and a CTO signing an enterprise contract are, functionally, looking at two different competitive landscapes with two different leaders.
From Monopoly to Marketplace: A Brief History of ChatGPT’s Lead
When ChatGPT launched in November 2022, it didn’t just lead the AI chatbot category. For most of the next three years, it effectively was the category. Momentic’s historical data put ChatGPT’s worldwide web-visit share as high as 79% in mid-2025. Apptopia’s app-market numbers, cited by Fortune, still showed ChatGPT at 69.1% as recently as January 2025.
2022-2025: The Uncontested Era
Google’s Bard, and later Gemini, struggled through multiple rebrands and underwhelming early releases. Anthropic spent that period focused on developer and enterprise tooling rather than consumer scale. Meta, xAI and DeepSeek were either late to the category or, in DeepSeek’s case, arrived with a single viral moment in early 2025 that didn’t translate into sustained share. ChatGPT’s first-mover advantage looked less like a lead and more like a monopoly, a framing one widely shared developer analysis on dev.to made explicit, arguing that the “monopoly phase” of AI assistants had defined the category since launch.
2026: The Erosion Becomes a Trend
That changed fast once it changed. FirstPageSage’s monthly tracking, shown in the table below, shows ChatGPT’s web-traffic share falling from 67.1% in January 2026 to 52.7% by July, a 14.4 percentage-point drop in six months. Sensor Tower’s app data shows a similar acceleration, with ChatGPT crossing below the 50% threshold sometime between January and the end of May. Whichever dataset is used, 2026 is the year the ChatGPT-only era of AI chatbots gave way to a genuine multi-player market.
What’s Driving the Fragmentation of the AI Chatbot Market
A few forces show up across nearly every report on this shift. The first is model quality convergence: Google’s Gemini 2.5 Pro with Deep Think reportedly scored 82.4% on GPQA Diamond after its June 22, 2026 launch, closing a gap that used to favor OpenAI by a wide margin. When a free or cheap alternative scores close to the category leader on independent benchmarks, price and distribution start to matter more than brand loyalty.
The second is distribution, covered above: Gemini’s default placement across Android and Workspace does work that marketing budgets cannot buy. The third is less visible and more human: a talent migration between labs. By late June 2026, Anthropic had reportedly hired two more senior Gemini researchers in a single week, the fourth senior departure from Google to Anthropic in six days, according to a roundup published by BuildFastWithAI’s June 26 news summary. Research talent moving toward a rival doesn’t show up in a market-share chart immediately, but it tends to show up in product quality a few quarters later.
Finally, OpenAI itself is spending part of 2026 on infrastructure rather than pure model iteration, including revealing its first custom AI chip this year, a bet on long-term cost control rather than a near-term share grab.
ChatGPT vs Gemini vs Claude vs Grok vs DeepSeek: Market Share At a Glance
Because no two trackers agree on methodology, the fairest way to compare the field is side by side, with each column clearly labeled by source. The table below lines up Sensor Tower’s app-market numbers against two separate web-traffic reads. Treat each column as internally consistent, and be cautious comparing figures across columns.
| Platform | Sensor Tower App Share (end-May 2026) | StatCounter Web Traffic Share (June 2026) | Similarweb-Based Web Share (May 2026, Momentic) |
|---|---|---|---|
| ChatGPT (OpenAI) | 46.4% | 76.87% | 53.9% |
| Gemini (Google) | 27.7% | 7.94% | 27.4% |
| Claude (Anthropic) | 10.3% | 3.74% | 8.2% |
| Perplexity | Under 5%* | 7.91% | N/A |
| Microsoft Copilot | N/A | 3.49% | N/A |
| DeepSeek | Under 5%* | 0.03% | 4.1% |
| Grok (xAI) | Under 5%* | N/A | 2.8% |
*Sensor Tower grouped Grok, Perplexity, DeepSeek and Meta AI collectively as each under 5%, without individual breakouts, per TechCrunch’s reporting. Sources: Sensor Tower State of AI 2026 (via TechCrunch), StatCounter Global Stats, Similarweb data via Momentic Marketing. Notice how the ranking order itself changes by column. Perplexity edges out Claude on StatCounter’s web numbers but trails badly on Sensor Tower’s app numbers, and DeepSeek’s share swings from statistically invisible to a real mid-single-digit player depending on the source. That instability is itself a data point about how unsettled this market’s measurement standards still are.
Market Share Trend: How Fast ChatGPT’s Lead Eroded in 2026
Zooming out from any single snapshot, FirstPageSage’s month-by-month tracking of ChatGPT’s web-traffic share shows how compressed this decline has been. The reports show a steady, close to linear slide across the first seven months of 2026, without a single month of recovery.
| Month (2026) | ChatGPT Web Traffic Share |
|---|---|
| January | 67.1% |
| February | 65.7% |
| March | 62.6% |
| April | 61.8% |
| May | 59.1% |
| June | 58.6% |
| July | 52.7% |
Source: FirstPageSage, “Top Generative AI Chatbots by Market Share.” A 14.4-point drop in six months would be a striking move for a company with ChatGPT’s scale in any industry, and First Page Sage’s separately reported monthly market-share read put the July figure closer to 51.3%, a reminder that even one firm’s own numbers shift slightly depending on exactly which tracker within its report is being cited. That it happened while ChatGPT was simultaneously setting usage records, 900 million weekly users in February, a figure QuickSEO’s 2026 market-data roundup was still citing as current later in the year, and the fastest run to 1 billion monthly users of any app in history by June, per Sensor Tower, shows that this is a story about relative position, not absolute decline. ChatGPT is not shrinking. Its rivals are growing faster.
Market Impact: What Slipping Share Means for OpenAI’s Next Moves
A ten-point or larger swing in market share, however it is measured, changes incentives, especially set against OpenAI’s own revenue base: Business of Apps put ChatGPT’s 2025 revenue at roughly $8 billion, a number that raises the stakes of defending share rather than shrugging off the slide. OpenAI has already begun monetizing its consumer base more aggressively. The company’s ad rollout inside ChatGPT responses reached 51% of US replies and went live in six countries earlier this year, a shift we covered in detail in ChatGPT Ads Hit 51% of US Replies. That timing is not a coincidence. When user growth alone can no longer be trusted to carry revenue growth, and rivals are closing the model-quality gap for free or near-free, advertising becomes a more urgent lever.
Expect pricing moves too. The AI assistant market in late 2025 and early 2026 already saw aggressive undercutting, with some providers pushing non-reasoning model pricing down toward $20 per million tokens against premium frontier offerings running closer to $500. If Gemini and Claude keep converting free distribution and enterprise wins into share, OpenAI’s options are to compete harder on price, differentiate further on agentic and coding features, or lean on its remaining scale advantage in consumer brand recognition. Reporting through mid-2026 suggests OpenAI is trying some combination of all three rather than picking one lane, and that split focus is itself part of why competitors have found room to grow.
5 Predictions for the AI Chatbot Market Through 2027
- ChatGPT’s app-market share keeps drifting lower, likely toward the high 30s by early 2027, even as its raw user base keeps growing, because the overall AI chatbot market is expanding faster than any single player can capture.
- Gemini closes in on Claude’s enterprise numbers as Google pushes harder into Workspace-bundled agentic tools, though Anthropic likely keeps its enterprise win-rate lead through at least the first half of 2027.
- At least one more mid-tier player breaks into consistent double-digit share on at least one major tracker, whether Perplexity, Grok, or a Chinese lab such as DeepSeek or MiniMax, ending the current three-horse framing.
- Disputes over whose market-share numbers are “real” get louder, as Sensor Tower, Similarweb, StatCounter and company-reported figures keep diverging, pushing analysts toward blended methodologies rather than single-source reports.
- OpenAI leans further into advertising and enterprise agent products rather than pure chat to defend revenue as consumer share keeps fragmenting, following the path already visible in its 2026 ad rollout.
What the Shift Means for Developers and Businesses Choosing a Platform
For engineering teams and businesses picking a primary AI vendor, the practical takeaway is not that ChatGPT is failing. It’s that there is no longer a single obvious default. A year ago, choosing ChatGPT for a new project was close to a safe, unexamined choice. Today, the right model depends heavily on the job: coding-heavy workflows increasingly favor Claude or DeepSeek-class models on cost and benchmark performance (see our best AI model for coding comparison for current numbers), agentic and Workspace-integrated tasks favor Gemini, and general consumer-facing products still lean toward ChatGPT’s brand recognition and scale.
That fragmentation is a genuine planning problem. Multi-model strategies, once a hedge for the risk-averse, are becoming close to standard practice, and procurement teams increasingly run side-by-side evaluations rather than defaulting to whichever provider got there first. Enterprise deals that would have automatically gone to OpenAI two years ago are, per the enterprise-tracking numbers above, now genuinely competitive. Software teams building AI features should treat model selection as a decision to revisit every quarter, not a one-time platform choice, given how quickly the underlying share and benchmark numbers keep moving.
FAQ: ChatGPT Market Share Questions Answered
What is ChatGPT’s market share as of mid-2026?
It depends on the tracker. Sensor Tower’s app-usage data put ChatGPT at 46.4% of the global AI-assistant market at the end of May 2026, the first time it fell below 50%. Web-traffic trackers show different numbers: StatCounter had ChatGPT at 76.87% of chatbot web visits in June 2026, while a Similarweb-based read from Momentic Marketing put it at 53.9% over roughly the same period.
Why did ChatGPT’s market share fall below 50%?
Analysts point to three main factors: Google’s Gemini gaining share through default placement on Android and Workspace, Anthropic’s Claude winning a majority of head-to-head enterprise deals, and improving model quality among competitors narrowing the performance gap that once made ChatGPT the obvious default choice.
How much share does Google Gemini have now?
Sensor Tower put Gemini at 27.7% of the global app market at the end of May 2026, up from roughly 14.7% in January 2025 per Apptopia’s data. Similarweb-based tracking from SudoFlare showed a similar 27.4% web-visit share in June 2026.
Is Claude growing faster than Gemini?
In percentage-growth terms, yes. SudoFlare’s data shows Claude’s monthly web visits growing 306% in a single quarter, from 203 million to 824 million. In absolute share, Claude still trails Gemini, sitting at roughly 10.3% of the app market per Sensor Tower.
Why do different reports show such different numbers for the same companies?
Each firm measures something different. Sensor Tower tracks app installs and sessions on mobile and desktop. StatCounter and Similarweb track visits to web domains. Regional coverage, bot filtering and which products count as “AI chatbots” all vary by provider, which is why DeepSeek’s share, for example, ranges from 0.03% to 4.1% depending on the source.
Is ChatGPT actually losing users?
No. ChatGPT’s absolute user base kept growing through mid-2026, crossing 900 million weekly active users in February and becoming, per Sensor Tower, the fastest app in history to reach 1 billion monthly users by June. It is losing share of a fast-growing market, not shrinking in absolute terms.
Which AI company is winning enterprise deals?
Anthropic’s Claude reportedly wins about 70% of new enterprise deals when businesses choose directly between Claude and ChatGPT, according to data cited by EnterpriseDNA. That is a different measure from consumer market share, where ChatGPT still leads by a wide margin in most reports.
Will ChatGPT’s market share keep falling?
Every tracker’s trendline through mid-2026 points down, with FirstPageSage showing a 14.4 percentage-point drop between January and July alone. Whether that slide continues, stabilizes or reverses through 2027 will likely depend on OpenAI’s next model releases and how it responds on pricing.
Related Coverage
- Sonnet 5 vs GPT-5.6 vs Gemini 3.1 Pro: 58% Cheaper [2026]
- Opus 4.8 vs GPT-5.6 vs Gemini 3.1 Pro: $18 Price Gap [2026]
- Claude Sonnet 5 Debuts: 57 Score, Half the API Cost [2026]
- Grok 4.3 vs Gemini 3.1 Pro: 4x Output Price Gap [2026]
- ChatGPT Ads Hit 51% of US Replies, Live in 6 Nations [2026]
- Best AI Model for Coding: DeepSeek Costs 20x Less [2026]
