Google has reduced the number of free image prompts available for its popular AI image generator, Nano Banana, from three to two. The move comes as the company struggles to manage soaring demand for its next-generation AI-enhanced photos. The change, first noted by 9to5Google, reflects Google’s efforts to balance accessibility with server capacity. The company reserves the right to adjust these limits further, citing fluctuating traffic volume.
Growing Constraints on Free Users
Alongside the Nano Banana adjustment, Google is also restricting free access to its Gemini 3 Pro model. Non-subscribers will now be relegated to “basic access,” where daily prompt limits may be capped at any time. Subscribers to Google AI Pro or AI Ultra plans are unaffected. This tiered system mirrors similar restrictions implemented by OpenAI for its Sora video generator, which now limits free users to six AI-generated video prompts per day.
“Our GPUs are melting, and we want to let as many people access Sora as possible,” stated Bill Peebles, head of Sora, on X.
Rapid Growth and Integration of Nano Banana
Nano Banana gained prominence after achieving high rankings on the LMArena AI leaderboard in August. A recent upgrade to Gemini 3 further boosted its capabilities, leading to its rapid integration across Google’s product suite. The AI tool is now embedded in Google Search, NotebookLM, Google Photos, and Google Messages. Google plans to extend Nano Banana’s reach even further, including direct integration into Android’s Chrome Canary search bar.
This expansion aligns with Google’s broader strategy to infuse its products with AI-powered assistance. Gemini 3 represents the company’s most advanced reasoning model to date, and was the first Gemini AI upgrade to be rolled out across all Google products. The initial launch granted users five free prompts per day; however, the rising popularity of the Gemini app (with over 650 million monthly users) has necessitated these adjustments.
The Bigger Picture
The scaling back of free access highlights the real-world constraints of AI infrastructure. While generative AI models are becoming increasingly sophisticated, the computational resources required to run them at scale are substantial. Limiting free usage ensures that Google can maintain service quality for paying subscribers while still allowing a degree of public access. This trade-off is becoming a common theme as AI adoption surges, and it raises questions about the long-term sustainability of open access in the face of exponential demand.
