Both Google and OpenAI have in fact decided to significantly reduce usage limits for their flagship models, Nano Banana Pro and Sora, citing global demand that has reached unsustainable levels for current hardware infrastructures.
Bill Peebles, Sora project lead at OpenAI, confirmed that users of the free plan will have a maximum of six video generations available per day. The justification provided was technical but eloquent: “Our GPUs are melting,” the executive explained, illustrating the computational stress that data centers are subjected to.
Unlike similar restrictions introduced in the past, on this occasion it was not specified whether the measure is temporary; it was instead underlined that users will be able to purchase additional generations when needed, a move that is clearly part of a broader monetization strategy for the platform. For subscribers to the ChatGPT Plus and Pro services, the limits remain unchanged for now.
At the same time, Google also imposed stricter restrictions. Non-paying users are now limited to creating just two images per day on Nano Banana Pro, a tool launched just last week. The downward change, compared to the previous limit of three images, was justified by the company: these thresholds are subject to frequent changes and “without notice”, a practice that is becoming standard after releases of great media attention. It also seems that the limitations are starting to include free access to the Gemini 3 Pro model.