The open bar period with the Google's most powerful AI, Gemini 3 ProIt didn't last long. The company quietly introduced a significant reduction in what free accounts can do, adjusting both the number of queries and the image creation capabilities of its Nano Banana Pro model.
What until recently was a relatively generous access system is now being replaced by a much stricter one. Google speaks of a “basic access” for non-subscribed userswhere the limits cease to be clear and fixed, depending instead on the level of platform usage at any given time, something that is already becoming noticeable among those who used Nano Banana Pro daily.
Goodbye to fixed limits: “basic” and variable access is here
Until now, free accounts had a fairly clear framework: they were offered five daily queries in reasoning mode of Gemini 3 Pro and the possibility of generating three images a day with Nano Banana ProThat was the deal for anyone who wanted to try the model without paying.
With the update to the support page, Google has removed that guarantee and replaced it with a system in which non-subscribed users only have a Limited access and subject to server loadIn other words, there is no longer a guaranteed minimum number: if the service is in high demand, the limit will be reached sooner and the user will be left out without much explanation.
This new framework means that the free version becomes, in practice, a congestion-conditioned test windowThe more people using Gemini 3 Pro at any given time, the less real margin those still on the free version will have, something that is already starting to be seen with occasional errors and crashes.
The company justifies the move by appealing to high demand and the need to protect the infrastructureThe surge in usage following the official launch of Gemini 3 and its associated image model has put the servers to the test, forcing them to cut costs where it is "cheapest": free access.
Image cropping: Nano Banana Pro, from three to just two a day
If there's one area where the cutbacks are especially noticeable, it's in the visual aspect. Google's help page already reflects that the Image generation with Nano Banana Pro is now limited to two creations per day. For free accounts, that is, one less image per day compared to the initial plan.
The company explains that this cut is in response to the “High demand” for editing and visual creation skillsThis is not an isolated movement in the sector: something similar happened when ChatGPT launched its famous Ghibli style to transform photographs, an experiment that ended up forcing OpenAI to stop the images for free accounts because, in the words of Sam Altman, the system was "melting graphics cards".
In Google's case, the pattern is repeated. The initial push to get everyone to try the new features has led to a bottleneck affecting computing resources. The chosen solution is clear: reduce what non-payers can do, maintaining the appeal of the features but greatly limiting how much they can be used without going through a paid plan.
Even with the new limits, the image model is still present in the free version, but no longer as a tool to use without looking at the counter, but as a small daily “appetizer” of what Nano Banana Pro offers when there are no restrictions.
Paid Gemini 3 Pro: More messages, more context, and more images
As the free experience shrinks, the paid version of Gemini 3 Pro It is becoming established as the way to work without so many restrictions. The subscription, with a price that is around 20 monthly dollars In the markets where it is available, it provides access to a much more comprehensive package for those who use AI intensively.
With this plan, users get Up to 100 messages per day with Gemini 3 Pro, a conversational context approximately four times larger than at the basic level and the option to generate 100 images per day with Nano Banana ProIn other words, a significant leap in both volume and capacity to work on long projects.
This twist reinforces the idea that the The most complete and advanced version of the model is reserved for those who pay a monthly fee.while the rest operate in an increasingly limited environment, with reduced functions and no clear guarantees for daily use.
For many users who have become accustomed to using Gemini 3 as their central tool for studying, working, or creating content, the change means being forced to seriously consider switching to a subscriptionespecially if they rely on AI for ongoing tasks and not just for one-off queries.
Google AI Plus and the push towards subscriptions
Alongside these cuts, Google has deployed Google AI PlusA more affordable subscription plan that aims to bridge the gap between the limited free access and more expensive options like AI Pro or Ultra. This service is priced around 7,99 euros per month, a figure designed for those who find the free version insufficient, but don't need the top of the range.
The appearance of this intermediate tier coincides with the reduction of privileges for basic accounts, making it difficult not to see a... clear strategy to push part of the user base towards paymentReducing free access, offering an "affordable" option, and keeping the most powerful plans intact is a familiar formula in digital services.
In Europe and Spain, where the monthly subscription model is widely established in video, music and software platforms, Google seems to be betting on a similar approach with its AI: a moderate fee in exchange for a complete experienceleaving free use as an occasional resource or to test the system before deciding.
For those who work in professional, educational or creative environmentsThis type of plan can become another cost in your digital tools "pack," in the same way as other services you already pay for on a monthly basis.
The impact on the ecosystem: NotebookLM and other features cut
The change of direction isn't limited to the Gemini 3 Pro chat. According to the official documentation, other pieces of the Google ecosystem They have also seen how features for free accounts are being shut down., at least temporarily while the servers are being reinforced.
Google insists that these decisions are in response to “severe capacity restrictions” There has been a continued increase in the user base. In fact, the congestion is so severe that, at certain times, problems have been reported even among people with paid plans, something the company implicitly acknowledges by promising to restore normal service "when possible."
All of this paints a picture where free access to Google's AI becomes a reduced version and subject to availabilitywhile the entire ecosystem, without cuts, is clearly aligning with the payment methods.
From open testing phase to limited service
Following its launch, the company opened a extensive trial phase for millions of peopleallowing users to utilize these capabilities at no cost for the first few weeks. This was a way to test the model, gather real-world usage data, and, incidentally, generate excitement around Gemini 3.
That phase, however, has reached its limit. Google has confirmed that for basic accounts, access is now restricted to a limited version of the mode of reasoningwith less margin for daily use and with limits that will no longer be predictable, since they will vary depending on how many people are connected at the same time.
The result is a fairly clear change of stage: from a period of almost unchecked testing to a service in which the More advanced AI is reserved for those who pay.And what remains for the rest is a limited experience, valid for occasional use but increasingly less suitable for regular tasks.
With these adjustments, Google draws a fairly clear line between what it offers for free and what it really wants to monetize. The Gemini 3 Pro and Nano Banana Pro remain available to everyone, but intensive use now depends on a subscription, while the basic level It becomes more unpredictable and restricted. For users in Spain and the rest of Europe who already integrated this AI into their daily lives, the current situation forces them to decide whether it's worth paying a monthly fee or lowering their expectations and frequency of use.
