Artificial intelligence (AI) image processing models are becoming a major driver of mobile application growth, beyond their previous role as chatbot upgrades. That’s the remarkable conclusion of a new report from Appfigures, a US-based company that provides a data analytics platform for mobile applications.
Data shows that the release of the new image creation model increased the number of AI application downloads by 6.5x compared to the previous model upgrade.
This trend shows that users are increasingly attracted to the ability to create visual content rather than just textual interactions and conversations.
Previously, advancements in chatbots, such as improved conversational capabilities and the addition of voice interfaces, were the main factors attracting new users. But things are rapidly changing as the AI that creates photos brings about intuitive, shareable, and more intriguing experiences.
Specifically, ChatGPT and Google Gemini both recorded tens of millions of new downloads after publishing their image models.
With the introduction of the Nano Banana image model and Flash Gemini 2.5 version, Gemini saw an increase in application downloads of over 22 million in just 28 days, more than four times the previous number.
Meanwhile, ChatGPT reached more than 12 million installations after the release of the GPT-4o image model, approximately 4.5 times more than previous versions such as GPT-5 and GPT-5. These numbers demonstrate the strong appeal of AI images in attracting new users.
A similar trend appeared when the Meta AI platform launched its AI video feature Vibes, which contributed to an increase of approximately 2.6 million downloads in 28 days. Although this product is a video model, it emphasizes the video experience, contributing to the strengthening of the trend toward emphasis on visual content.
However, the report also points out the inconsistency that an increase in downloads does not mean an increase in revenue. Many users download the application to try out the photo creation features, but do not convert into paid users.
For example, Gemini’s Nano Banana model only generated about $181,000 in revenue over 28 days, despite a sudden increase in load. Similarly, Meta AI recorded an increase in load but did not generate significant revenue.
Among the major platforms, only ChatGPT demonstrated effective conversion capabilities. The GPT-4o image creation model helped OpenAI increase consumer revenue by approximately $70 million in the first 28 days of launch.
The report also mentions DeepSeek as a special case. The DeepSeek R1 model has attracted up to 28 million downloads, which is believed to be a boom effect due to market curiosity about low-cost technology rather than a trend in AI imaging.
In general, the data shows that AI is entering a new phase and image experience will become a key factor in attracting users. But the problem of moving from downloads to revenue remains a major challenge for technology companies.
