Imagery AI models outpace chatbot upgrades to drive app growth

Image model releases are driving the growth of AI mobile apps, generating 6.5 times more downloads than traditional model updates, according to a new report from app intelligence provider Appfigures.

This marks a change from earlier times when demand increased with the release of new models that enhanced the conversation experience, along with new features such as voice chat interfaces.

For example, Appfigures found that ChatGPT and Gemini each added tens of millions of new downloads after releasing their respective image models.

For Google’s Gemini, the release of the image model Nano Banana led to over 22 million additional downloads in the 28 days since the introduction of the Gemini 2.5 Flash image model last August. Data shows that this release led to a more than 4x increase in app downloads during that period.

Meanwhile, ChatGPT added more than 12 million incremental installs in the 28 days since introducing the GPT-4o image model last March. Appfigures noted that this is approximately 4.5 times more downloads than the GPT-4o, GPT-4.5, and GPT-5 model releases.

Other model releases followed similar trends, but on a smaller scale. The introduction of Vibes, an AI video feed by Meta AI, added an estimated 2.6 million additional downloads in the 28 days since its September 2025 release. (Yes, technically this is a video model, but ultimately it’s about visual content, not just text.)

Still, the report cautions that additional downloads do not necessarily translate into increased mobile revenue.

Instead, releasing a new image model gives people a reason to install the app and try out improved image generation capabilities. That doesn’t necessarily mean converting to a paid membership. For example, Appfigures noted that Nano Banana generated a larger download spike than ChatGPT’s 4o image model release, yet generated only $181,000 in estimated total consumer spending in the 28 days following its release.

The launch of Vibes by Meta AI also led to additional downloads, but no meaningful revenue.

Of the three, only ChatGPT translated its increased attention into actual dollars.

According to Appfigures, OpenAI’s 4o image generation model generated an estimated $70 million in total consumer spending in the first 28 days of launch compared to the previous baseline.

The company also considered DeepSeek in its analysis, but it didn’t fit the pattern.

DeepSeek R1 racked up 28 million downloads after its release in January 2025, but this was not your typical model comparison event. This was DeepSeek’s breakthrough moment, going from relative obscurity to overnight sensation as the tech industry learned of the technology it uses to train AI models at a fraction of the cost of its competitors. This case highlights how curiosity drives downloads. However, in this example, the interest was not associated with the image model.