Visuals Over Verbiage: Image Models Surpass Chat Upgrades as AI App Growth Engines

AI image models: The competitive landscape of artificial intelligence has shifted from conversational prowess to visual creativity. According to a new report from app intelligence provider Appfigures, the release of image-generation models is now the primary catalyst for mobile app growth, significantly outperforming traditional chatbot updates. On average, image model launches generate 6.5x more downloads than standard text-based model upgrades.

This trend highlights a fundamental change in consumer behavior: while voice chat and text-based reasoning were the initial draws for AI, users are now flocking to platforms that offer high-fidelity visual content creation. Industry giants like Google and OpenAI have already capitalized on this shift, seeing massive surges in user acquisition following the introduction of their respective image-focused capabilities over the past year.

The Download Surge: Gemini and ChatGPT

The data reveals a stark contrast between the impact of “brain” upgrades versus “vision” upgrades. Google’s Gemini experienced a massive spike following the release of its Nano Banana image model. In the 28 days following the introduction of the Gemini 2.5 Flash image model last August, the app saw an additional 22 million downloads. This launch alone lifted the app’s installation rate by more than 4x compared to previous periods.

OpenAI’s ChatGPT followed a similar trajectory. When the GPT-4o image model was introduced in March 2025, it triggered over 12 million incremental installs within a month. To put that in perspective, the image model release was roughly 4.5x more effective at driving new users than the releases of the GPT-4.5 or GPT-5 text models combined. Even Meta AI entered the fray with its Vibes video feed, which, while technically a video model, leveraged the demand for visual content to secure 2.6 million new downloads in late 2025.

The Revenue Gap: Downloads vs. Dollars

While image models are unparalleled at getting users through the door, the report highlights a significant challenge: monetization. A massive spike in downloads does not automatically equate to a healthy bottom line.

• Google Gemini: Despite the 22 million new users brought in by Nano Banana, the model generated only an estimated $181,000 in gross consumer spending during its first month.

• Meta AI: The Vibes launch resulted in virtually no meaningful revenue increase, as the platform remains largely focused on engagement within the broader Meta ecosystem.

• OpenAI: Standing as the outlier, ChatGPT successfully converted curiosity into capital. The GPT-4o image model led to approximately $70 million in gross consumer spending over its launch window.

This suggests that while users are eager to try “free” image tools, OpenAI has been more successful in locking those users into a subscription-based ecosystem.

The DeepSeek Outlier

The report also touched on DeepSeek, which broke the pattern in early 2025. While its R1 model drove 28 million downloads, this growth wasn’t fueled by image generation. Instead, it was driven by industry-wide curiosity regarding DeepSeek’s ability to train high-level models at a fraction of the cost of Western competitors. This “breakout moment” proves that while visuals are the current trend, fundamental tech disruption can still move the needle.

Summary of Impact

Ultimately, the AI image industry is entering a “visual-first” era. Developers are finding that while text models provide the utility, image and video models provide the “wow factor” necessary to dominate the crowded app store charts. Moving forward, the industry’s hurdle will be finding ways to make these resource-heavy visual tools as profitable as they are popular.

Also read: UAE Targets AI-Run Government in 2 Years: Global Race Intensifies