Image AI models surpass chatbots in driving app growth

The landscape of AI mobile app growth is undergoing a significant transformation, with image model releases now driving far more downloads than traditional chatbot updates. According to a recent analysis from app intelligence firm Appfigures, these visual-focused launches generate 6.5 times more installations than standard model refreshes.

This represents a clear departure from the early days of AI apps, when updates to conversational features or voice interfaces were the primary growth catalysts. Today, the ability to generate and manipulate images has become a powerful magnet for users.

Concrete examples illustrate this trend. Google’s Gemini saw a massive surge after releasing its Nano Banana image model as part of the Gemini 2.5 Flash update last August. In the 28 days following that launch, the app recorded over 22 million additional downloads, representing a 4x increase in its normal acquisition rate.

Similarly, OpenAI’s ChatGPT experienced a substantial boost when it introduced its GPT-4o image model in March of last year. That release generated more than 12 million incremental installs in the subsequent four weeks. Appfigures notes this was roughly 4.5 times more downloads than what ChatGPT saw from its combined GPT-4o, GPT-4.5, and GPT-5 model launches.

Smaller players have also benefited. Meta AI’s introduction of Vibes, an AI video feed, added an estimated 2.6 million incremental downloads in the 28 days after its September 2025 release. While technically a video model, it underscores the broader appeal of visual content over pure text.

However, Appfigures cautions that a spike in downloads does not automatically translate into higher mobile revenue. The data reveals a stark disconnect between user acquisition and monetization.

For example, Nano Banana drove only an estimated $181,000 in gross consumer spending during its 28-day post-launch window, despite generating a larger download spike than ChatGPT’s 4o image model. Meta AI’s Vibes also brought in new users but yielded no meaningful revenue.

Among the three major players, only ChatGPT successfully converted the increased attention into significant financial returns. OpenAI’s 4o image-generation model led to an estimated $70 million in gross consumer spending over the 28 days after its release, compared to its prior baseline.

The report also examined DeepSeek, but its trajectory did not follow the same pattern. When DeepSeek R1 launched in January 2025, it drove 28 million downloads. However, this was not a typical model comparison event. Instead, it was DeepSeek’s breakout moment, as the tech world became fascinated by its ability to train AI models at a fraction of the cost of competitors. This case highlights how curiosity and novelty can drive massive downloads, even when the appeal is not tied to an image model.