ChatGPT might’ve been the first mover in the AI space, but Google’s Gemini is slowly and steadily catching up.
Google’s Gemini has overtaken OpenAI’s ChatGPT in a key engagement metric: average time spent per visit on desktop and mobile web platforms. According to recent data by FT, Gemini users now spend approximately 7.2 minutes per session, surpassing both ChatGPT at around 6 minutes and Claude at roughly 6 minutes as of late 2025.
The shift represents a significant milestone in the competitive AI chatbot landscape. When ChatGPT launched in late 2022, it became an immediate phenomenon, reaching 100 million users faster than any consumer application in history. OpenAI’s early dominance seemed unassailable, with the platform maintaining strong engagement metrics through 2024 and into early 2025, peaking at around 7 minutes per visit in April 2025.
However, Gemini’s trajectory tells a different story. After struggling with lower engagement in mid-2024—dipping to around 5 minutes per visit—Google’s AI assistant has experienced a dramatic resurgence. The platform’s recent developments, including the release of the Gemini 3 Pro model and Nano Banana Pro earlier this month, appear to be resonating with users who are spending more time exploring its capabilities.
The app download numbers paint a similar picture of Gemini’s momentum. While ChatGPT maintains a substantial lead with approximately 87 million monthly downloads as of late 2025, Gemini has shown explosive growth, jumping from roughly 15 million monthly downloads in mid-2025 to around 66 million by late 2025. This surge suggests that Google’s integration of Gemini across its ecosystem—from Android devices to Google Workspace—is paying dividends in user acquisition.
What makes the time-per-visit metric particularly revealing is that it reflects genuine user engagement rather than just curiosity-driven downloads. Users are choosing to spend more time with Gemini for their queries, suggesting that the platform’s responses, features, or overall experience are compelling enough to hold attention in an increasingly crowded market.
The competitive dynamics extend beyond just ChatGPT and Gemini. Anthropic’s Claude has maintained relatively stable engagement around 6 minutes per visit, while other players like Perplexity and DeepSeek remain significantly smaller in terms of app downloads, though they serve niche audiences with specialized capabilities.
For Google, the turnaround is particularly sweet given the company’s initial stumbles with AI chatbot launches. The early version of Bard, Gemini’s predecessor, faced criticism for inaccuracies and lackluster performance compared to ChatGPT. The current engagement figures suggest that Google has successfully addressed many of those concerns while leveraging its vast computational resources and integration opportunities, including the Nano Banana image generator, and its latest Gemini 3 Pro model.
The battle for AI dominance is far from over. OpenAI continues to innovate rapidly, and ChatGPT’s first-mover advantage still provides substantial brand recognition and user loyalty. But Gemini’s recent performance demonstrates that the AI race remains wide open, with user preferences still fluid and competitors just one breakthrough model away from shifting the balance of power once again.