🥇 ChatGPT tops global downloads and reshapes the race for AI dominance

🚀 OpenAI rolls out memory and ID verification to strengthen its leadership 📱 ChatGPT hits 46 million downloads and overtakes Instagram on iOS

Artificial intelligence doesn’t take breaks — and neither does Sam Altman. During the 2025 Holy Week, ChatGPT became the most downloaded app on the planet. For the first time since 2013, a non-social media app leads the global download rankings. According to Appfigures, ChatGPT reached 46 million installs in March, beating out giants like Instagram, TikTok, and WhatsApp.

On iOS devices, ChatGPT leads with 13 million downloads, while on Android it ranks third behind Instagram and TikTok, with 33 million. This trend reflects how Apple users are turning to GPT to get what Siri still doesn’t offer. The gradual integration of OpenAI’s model with Apple’s voice assistant is changing the rules of the game.

March’s success can’t be explained without the viral phenomenon of Studio Ghibli-style images. Opening up the image generator to free users triggered a surge of activity. In just one hour, five million new users joined the platform. GPUs were overheating, and Altman asked for patience: “Our team needs sleep.”

But beyond virality, OpenAI is solidifying a long-term leadership strategy. The company has launched a new series of models: GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano — all available via its API for developers. These models are not yet part of the ChatGPT app but are optimized for programming, long-context understanding, and accurate instruction-following.

GPT-4.1 boasts a one-million-token context window, significant improvements over GPT-4.5, and top-tier performance in benchmarks like SWE-bench and Video-MME. With scalable pricing — from $0.10 per 1,000 tokens up to $8 for full models — OpenAI aims to meet a wide range of development needs.

Still, a major challenge remains: monetization. Despite its global popularity, ChatGPT is not profitable. Running GPT-4 comes at a high cost, and most users still rely on the free version powered by GPT-3.5. With 10 million Pro subscribers and another million on enterprise plans, revenue still falls short of covering expenses. Altman has admitted that paying users “use it way more than expected,” putting the service’s sustainability at risk.

The ongoing trade war with China and the rising costs of data centers could further increase the price of AI access. Aisera, a company in the sector, warns that these systems may become unsustainable unless a better balance is struck between free and paid usage. Comparisons to Google Gemini are unavoidable: its models are cheaper and equally effective in many benchmarks.

In response, OpenAI has introduced an identity verification system for API access. Aimed at developers, the move seeks to prevent misuse of AI and strengthen security across the tech ecosystem. It will be a mandatory requirement for accessing the platform’s advanced features. For now, it doesn’t affect general ChatGPT users — only those integrating the technology into other services.

Another key update is persistent memory. Available for Pro subscribers, ChatGPT can now remember previous conversations to tailor replies based on context. This memory can be customized, manually deleted, and alerts users when prior data is used. However, it’s currently unavailable in the EU and several other European countries.

OpenAI has also introduced a “temporary chat mode” for users who prefer not to leave a trace in the model’s memory. Ideal for more private queries, this option ensures that no data is stored. Any user can review and delete stored data via their settings panel.

These updates come alongside new creative tools in the app. Users can now generate images from text, transform personal photos into artistic styles, and produce near-professional results without any technical knowledge. These features have been key to onboarding a new wave of users who once saw AI as distant or irrelevant.

Altman believes that this ability to create and convey visual emotion from a text-based assistant could be one of the key catalysts for the digital age. “We want ChatGPT to be more than an assistant — we want it to be a tool for emotional connection and personal creativity,” he says.

Moreover, OpenAI has teased two more model launches for later this year: o3 and o4-mini. Although there were initial plans to merge capabilities into a single model, market complexity has led to keeping them as separate releases.

Meanwhile, other players in the sector are making moves. Midjourney, a benchmark in AI image generation, has launched its V7 model after a year without updates. Despite its technical quality, growth has slowed due to a less accessible interface and a continued focus on Discord — far from OpenAI’s seamless integration.

Freepik’s CEO has criticized the EU for focusing solely on the risks of AI, stating that “innovation doesn’t stop.” His company is pushing to embed creative AI models into mainstream platforms, emphasizing user experience and mass adoption.

Startups are also stirring. Safe Superintelligence, founded by Ilya Sutskever, has raised $2 billion without having a public product yet. Valued at $32 billion, SSI promises to develop AI with “nuclear-grade safety.” Another OpenAI alum, Mira Murati, is also preparing a multi-billion-dollar funding round for Thinking Machines Lab, supported by a team of former GPT leads.

These developments show that, despite lingering questions about profitability, faith in AI remains unwavering. The draw of powerful teams — more than specific products — has become a key driver in the venture capital world.

In short, ChatGPT has not only become the most downloaded app globally, it has kicked off a new era in the AI race. One where distribution, integration, and emotional resonance matter more than raw technical superiority. And as competitors scramble for position, Altman accelerates — convinced that the future is already here and the best AI will be the one that lives alongside millions of users: seamlessly, usefully, and meaningfully.