Image model launches are now driving stronger mobile app growth than traditional AI model updates, according to Appfigures data reported by TechCrunch. The report found that image model releases generated 6.5x more downloads than standard chatbot or language model upgrades.
TL;DR
- Image AI launches are giving users a stronger reason to install AI apps.
- Appfigures found Gemini and ChatGPT image models drove major download spikes.
- Revenue gains varied significantly across platforms.
Image Models Are Driving AI App Downloads
According to TechCrunch, Appfigures found that image model releases are now generating more mobile app growth than traditional AI model updates.
The report said image model launches produced 6.5x more downloads than standard model updates. This marks a shift from earlier AI app growth, when new conversational models and features such as voice chat were bigger drivers of demand.
Google’s Gemini saw one of the biggest spikes. Appfigures found that the release of Gemini 2.5 Flash Image, also known as Nano Banana, drove more than 22 million additional downloads in the 28 days after its launch in August 2025.
The launch lifted Gemini app downloads by more than 4x during that period. Google described Nano Banana as an image generation and editing upgrade in the Gemini app.
ChatGPT also saw a large increase after OpenAI introduced GPT-4o image generation in March 2025. Appfigures found that ChatGPT added more than 12 million incremental installs in the 28 days after the launch.
The report said that was roughly 4.5x more downloads than ChatGPT saw after its GPT-4o, GPT-4.5, and GPT-5 model releases. OpenAI described GPT-4o image generation as a native capability that can create photorealistic output and transform images.
Downloads Did Not Always Mean Revenue
TechCrunch also highlighted that higher downloads did not always lead to higher revenue.
Appfigures estimated that Nano Banana generated only $181,000 in gross consumer spending during the 28-day window after release, despite its strong download performance.
By comparison, OpenAI’s GPT-4o image-generation model led to an estimated $70 million in gross consumer spending over the same 28-day period.
Meta AI’s Vibes also produced additional downloads but did not generate meaningful revenue. Meta launched Vibes in September 2025 as a feature for creating and sharing short-form AI-generated videos.
Topics For More Insights
DeepSeek Was Treated As A Separate Case
The Appfigures analysis also examined DeepSeek, though TechCrunch noted it did not follow the same pattern.
DeepSeek R1 drove 28 million downloads after its January 2025 release. However, this growth was attributed to broader attention around its low-cost AI training approach rather than a standard feature update cycle.
Join The Discussion