ChatGPT’s Lead Erodes as Gemini and Claude Carve Out Lasting Ground

ChatGPT once commanded three-quarters of AI chatbot web visits. That figure now sits near half. The shift marks more than a simple dip in traffic. It signals how distribution, product quality and enterprise preferences reshape a market that looked like a runaway winner for OpenAI just 18 months ago.

Data from Similarweb tracked through May 2026 shows ChatGPT falling from 76 percent to 54 percent of worldwide AI-chatbot web visits. Magnitude via Similarweb put the exact drop at 22 points. Gemini climbed the same distance, from 6 percent to 28 percent. Claude rose from 1 percent to 9 percent. These numbers capture desktop and mobile browser activity. They exclude in-app usage that often favors Google’s integrated Gemini experience on Android devices.

Yet web traffic tells only part of the story. App-store analytics paint an even sharper picture of erosion. Sensor Tower’s State of AI 2026 report recorded ChatGPT’s share of the global AI-assistant app market at 46.4 percent by the end of May 2026. TechCrunch first highlighted the milestone in June. For the first time since its November 2022 launch, OpenAI’s product no longer held a majority. Gemini stood at 27.7 percent. Claude reached 10.3 percent. The rest, including xAI’s Grok, Perplexity, DeepSeek and Meta AI, each stayed under 5 percent.

OpenAI still claims massive scale. The company announced 900 million weekly active users in February 2026. It became the fastest app ever to hit 1 billion monthly users by June. Raw volume remains enormous. But competitors chip away at the dominance. Users now switch between tools. They pick Gemini for quick searches tied to Google services. They turn to Claude for longer, more careful writing and coding tasks. Fragmentation has set in.

First Page Sage’s July 2026 tracking reinforces the pattern. The firm measured ChatGPT at 51.3 percent for the month, down from 67.1 percent in January. Gemini had grown to 27.7 percent from 14 percent over the same stretch. Claude held 10.3 percent. First Page Sage noted the top three tools still account for nearly 90 percent of activity. The market shows signs of stabilizing in the 50-to-55 percent range for ChatGPT rather than collapsing further. Still, the direction is unmistakable.

Why the change? Distribution gives Google an edge few can match. Gemini sits as the default assistant across Workspace, Android and Search. That built-in presence drives usage without extra effort from users. Anthropic, meanwhile, wins on quality and focus. Enterprise buyers reportedly hand Claude roughly 70 percent of head-to-head evaluations, according to reports cited in industry analysis. Its models excel at careful reasoning and avoid some of the hallucination traps that plague earlier systems.

Model performance has converged enough that small differences matter less than they once did. Stanford’s 2026 AI Index Report captured the broader acceleration. Industry labs produced more than 90 percent of notable frontier models in 2025. Several now meet or exceed human baselines on PhD-level science questions, multimodal reasoning and competition mathematics. SWE-bench Verified scores jumped from 60 percent to nearly 100 percent in a single year. Capabilities keep rising fast. Stanford HAI described the progress as anything but plateaued.

Adoption tells another side of the expansion. Eighty-eight percent of organizations now use AI in at least one business function. Four out of five U.S. students rely on generative tools. Global consumer value from AI reached $172 billion in recent measurements. Yet public trust lags. Only 23 percent of the general population express strong confidence compared with 73 percent of experts. These gaps influence how companies position their products and how regulators respond.

Spending forecasts reflect confidence in continued growth. Gartner projected worldwide end-user spending on AI platforms and models would hit $64 billion in 2026. That represents a 63 percent jump from $39 billion the year before. The wider AI market shows even larger numbers. Fortune Business Insights expects the global artificial intelligence sector to expand from $376 billion in 2026 to $2.48 trillion by 2034 at a 26.6 percent compound annual growth rate. North America held a 32 percent share in 2025 and remains the largest region.

Web traffic data from other trackers adds nuance. Statcounter recorded ChatGPT at 77.9 percent of AI chatbot market share in July 2026, with Gemini at 9.9 percent, Perplexity at 5.9 percent and Claude at 3.2 percent. The variance comes down to methodology. Some count unique users. Others measure visits or sessions. Similarweb-based figures from Momentic and First Page Sage consistently show steeper declines for ChatGPT than pure Statcounter numbers. The truth sits somewhere between the extremes. ChatGPT still leads by a wide margin. Its once-overwhelming advantage has narrowed.

Enterprise dynamics complicate the picture further. Anthropic captured 40 percent of enterprise LLM API spend in recent quarters while OpenAI’s portion slipped from 50 percent to 27 percent, according to one analysis. Businesses value Claude’s focus on safety and reliability for high-stakes work. They integrate multiple models rather than rely on a single provider. That multi-model approach accelerates the fragmentation seen in consumer data.

Smaller players nibble at the edges. Perplexity built a loyal following for its search-focused answers with sources. DeepSeek and Chinese models gain traction in specific regions and use cases. Grok, from xAI, appears in the under-5 percent tier in most trackers but draws attention for its real-time knowledge and less filtered responses. None yet threaten the top three. Together they prevent any one company from reestablishing total control.

The numbers also hide shifts in user behavior. Many people now keep several AI tools open at once. They route different tasks to the system that handles them best. A student might ask Gemini for research summaries, Claude for essay outlines and ChatGPT for creative brainstorming. This mixing reduces any single tool’s share of total interactions even as overall usage climbs.

Investment continues to pour in. The United States led with $286 billion in AI funding in 2025, more than 23 times China’s total. Nearly 2,000 new AI companies launched in the U.S. that year. China counters with volume in publications, citations and robotics deployments. The Stanford report highlighted a closing gap in certain metrics even as the U.S. maintains leads in model quality, patents and data-center capacity.

Incidents rose alongside capability. The AI Index counted 362 notable AI-related events in 2025, up from 233 the prior year. Concerns over accuracy, bias and security influence buyer decisions. Companies that score higher on transparency and controllability win trust in boardrooms even if their consumer-facing products hold smaller shares.

So what comes next? ChatGPT’s absolute user numbers keep growing. Its brand remains the strongest in the category. But the era of one dominant platform appears over. Google’s distribution muscle, Anthropic’s enterprise momentum and the rapid pace of model improvement across the board have created a competitive field where share points move quickly.

Analysts expect the market to keep expanding. Gartner’s forecasts and broader industry projections point to trillions in value over the coming decade. Within that growth, the battle for attention will center on integration, specialization and trust. OpenAI must find ways to regain ground without simply copying rivals’ advantages. Google will look to convert its ecosystem reach into deeper engagement. Anthropic will try to turn its quality edge into broader consumer adoption.

The data through mid-2026 shows a clear trend. Dominance has given way to a more balanced contest. ChatGPT lost 22 points of web share in a year. Gemini and Claude took most of those points. The next year will reveal whether that redistribution stabilizes or accelerates as new models and features enter the market. One thing looks certain. The AI assistant space no longer belongs to a single name.