For the first time since generative AI entered the mainstream, the race is no longer a one-horse sprint. Google’s Gemini has caught up with OpenAI’s ChatGPT on distribution, usage, and capability. Even as the gap narrows, however, the jury is still out on who ultimately leads the AI era.
ChatGPT had what every category-defining product dreams of: first-mover advantage, explosive adoption, and near-total mindshare. It didn’t just launch early; it became synonymous with AI itself. But with Gemini 3 and a rapidly expanding user base, Google has shifted the narrative from playing catch-up to posing a genuine challenge.
Powered by models such as Gemini 2.5 Flash (including the Nano Banana image variant) and Pro, that momentum is being fuelled by deep integration across the Google ecosystem – Search, Gmail, and productivity suites – enabling smoother workflows, real-time information access, and stronger contextual relevance.
Data Shift
Gemini’s monthly active users increased by more than 30% between August and November this year, reaching 346 million, according to data from research firm Sensor Tower. It now holds over 50% of India’s download market share. On its third-quarter earnings call, Google placed Gemini’s monthly active users significantly higher, at 650 million.
ChatGPT continues to lead in absolute numbers, with 810 million total monthly active users, but its growth over the same period was just 5%. Gemini’s desktop visits doubled during this time, while ChatGPT’s rose by only 1%. Weekly active users grew 52% for Gemini and 18% for ChatGPT from the beginning of August, the Sensor Tower report found.
Features such as the viral image generator Nano Banana, along with the launch of Gemini 3, have accelerated adoption. At the beginning of 2025, Gemini had fewer than 100 million weekly active users, compared to well over 200 million for ChatGPT, a gap that has since narrowed sharply.
“Nano Banana enhances Gemini’s multi-modal capabilities through conversational image generation and editing, delivering precise, context-aware visuals that elevate creative and professional workflows,” said Prabhu Ram, vice president, Industry Research Group at CyberMedia Research (CMR). “This positions Gemini strongly for enterprise and research-oriented users who prioritise accuracy and efficiency over ChatGPT’s broader creative flair.”
ChatGPT’s dominance is also being tested by challengers such as Perplexity and Anthropic’s Claude. Both recorded triple-digit growth in 2025, with Perplexity expanding 370% year-on-year and Claude growing 190%. Analysts say models are improving so rapidly that being first may no longer matter as much as being best right now. As newer systems better understand intent and require fewer prompts, users are increasingly willing to switch platforms.
Ram said that Perplexity is strengthening its niche in citation-backed research, functioning as a precision search tool in contrast to legacy “needle-in-a-haystack” search models. Claude, particularly its 3.5 Sonnet and Opus variants, continues to lead in coding depth and ethical enterprise automation, making it attractive to regulated sectors where reliability, governance, and compliance outweigh conversational versatility.
Beyond Benchmarks
According to Jacob Joseph, vice president of data science at CleverTap, Gemini 3 has forced the industry to rethink how progress is measured – not just in terms of higher reasoning scores, but smoother workflows, better planning, and a more natural grasp of intent. A key driver of this shift is the rise of on-device AI.
“Google’s work with Gemini Nano and Banana Pro signals a different philosophy of intelligence,” Joseph said. “Instead of routing every action through the cloud, these models work locally, resulting in lower latency, stronger privacy, and usefulness even when connectivity drops.”
India is poised to become the global testbed for this transition. “We have an enormous spectrum of devices, a massive mobile-first population, and now a zero-switch-cost environment created by GPT Go being free, Gemini bundled with Jio, and Perplexity arriving through Airtel,” Joseph said. If a model can deliver intelligence instantly and reliably in India, he added, it can deliver it anywhere.
OpenAI’s reported internal “code red” reflects this shifting landscape. In a memo to staff, CEO Sam Altman urged teams to accelerate improvements across OpenAI’s products, particularly in personalisation, reliability, image generation, and overall user experience.
Arjun Nagulapally, CTO at AIONOS, said today’s AI models show far stronger alignment with user intent than even a year ago. “They are not just better in language fluency, but in context retention, reasoning scaffolding, and task abstraction,” he said. The new competitive dynamic, he added, is no longer just about performance. It is about ecosystem lock-in through habit formation and workflow intelligence. The market is moving from which model is best to which system knows me best.
Joseph believes the next decisive advantage will come from delivering intelligence that feels effortless, like systems that understand intent, tolerate imperfect phrasing, work without ceremony, and remain responsive regardless of network conditions. That is the battleground Gemini has entered, and it presents a real challenge. Can ChatGPT respond?
“Absolutely,” Joseph said, “but it will need to compete on distribution and usability, not just model depth. GPT-5.2 is clearly OpenAI’s attempt to close that gap.”
