AI image generators fuel app growth faster than chatbot updates

Image model releases are increasingly driving growth for AI mobile apps, delivering 6.5 times as many downloads as traditional model updates, according to a new report from app intelligence provider Appfigures.

This represents a shift from earlier phases of the AI boom, when updates to conversational models and features like voice chat interfaces were the main drivers of user adoption. Now, visual AI capabilities are taking centre stage.

For example, both ChatGPT and Google Gemini recorded tens of millions of additional downloads after rolling out their respective image-generation models, according to the report.

In the case of Gemini, the launch of its image model, Nano Banana, led to more than 22 million additional downloads within 28 days after the release of the Gemini 2.5 Flash image model last August. This rollout increased the app's downloads by more than 4x during that period, according to the data.

Meanwhile, ChatGPT saw an increase of more than 12 million installs within 28 days after introducing its GPT-4 image model in March of the previous year. According to Appfigures, this represents roughly 4.5 times as many downloads as the app achieved with its GPT-4o, GPT-4.5, and GPT-5 model updates.

Other launches showed similar, though smaller, effects. Meta AI introduced its AI-powered video feed Vibes, which contributed approximately 2.6 million additional downloads in the 28 days following its September 2025 debut. Although technically a video feature, it still falls within the broader category of visual AI experiences rather than text-based interactions.

However, the report notes that higher download numbers do not always translate into increased revenue. Image model launches tend to attract users interested in trying out new features, but that interest doesn't always convert into paid subscriptions.

For instance, Appfigures estimated that Nano Banana generated only $181,000 in gross consumer spending during the 28 days after launch, despite producing a larger spike in downloads than ChatGPT's 4o image model rollout. Similarly, Meta AI's Vibes feature boosted downloads but did not lead to meaningful revenue gains.

Among the platforms analysed, only ChatGPT successfully translated increased attention into significant revenue. OpenAI's 4o image-generation model resulted in an estimated $70 million in gross consumer spending over the 28 days following its release, compared to its previous baseline, according to the report.

The analysis also included DeepSeek R1, though its case differed from the broader trend. DeepSeek R1 drove 28 million downloads after its January 2025 launch, but this surge was tied to its breakout moment in the industry rather than a typical model update. The model gained widespread attention for demonstrating how AI systems could be trained at a fraction of competitors' costs, sparking significant curiosity-driven downloads rather than interest in visual features specifically.