Image AI models drive 6.5x more app downloads than chatbot upgrades

A recent analysis by Appfigures, a provider of app intelligence, indicates that the launch of image models has become the leading factor in the growth of AI mobile applications, producing 6.5 times the downloads compared to conventional model updates. This represents a change from previous trends where conversational models and voice chat enhancements fueled interest. Google's Gemini saw its image model, Nano Banana, released in August 2025, resulting in an additional 22 million downloads within 28 days, more than quadrupling its total. In March 2025, ChatGPT's introduction of its GPT-4o image model led to over 12 million new installs in 28 days, about 4.5 times more than its previous models. Meta AI's Vibes video feed added approximately 2.6 million downloads after launching in September 2025. However, the report warns that increased downloads do not necessarily equate to higher mobile revenue. Nano Banana generated an estimated $181,000 in gross consumer spending over 28 days, while Vibes produced no significant revenue. In contrast, ChatGPT's 4o model resulted in an estimated $70 million in gross consumer spending during the same period. Additionally, DeepSeek R1 achieved 28 million downloads following its January 2025 release, driven more by interest in its economical training methods than by its image model.