NanoCloud

Google Models Under More Than One Name

The homepage describes NanoCloud as a “visual model built on Gemini 3 Pro,” while simultaneously promoting “NanoCloud Vision 2 powered by Nano Banana.” The API documentation, however, is more precise and useful, as it lists the actual options: nano-banana-fast, nano-banana, nano-banana-pro, two versions labeled “VIP,” and nano-banana-2. “Vision 2,” then, is a marketing name layered over Google models, not a model trained by NanoCloud itself. There is nothing wrong with that as long as you know it; what you are paying for is the interface, credit packages, and software wrapper, not a different engine.

The site also includes pages for the Sora 2 video model from OpenAI and Grok Imagine from xAI, making it closer to a platform that aggregates multiple models than to a studio built around a single model.

What the Studio Does Well

The features are those of the Nano Banana models themselves, but in a well-organized wrapper:

• Text-to-image and image-to-image at 1K, 2K, or 4K resolution, with a choice of aspect ratio, file format, preset styles, and prompt templates.
• Reference image blending: the homepage mentions 14 images, while the API documentation specifies eight images per request, with each image not exceeding 10 MB. So rely on the lower number if you are building on the API.
• Semantic editing: replace an element, change the mood, or expand the canvas simply by describing it, while maintaining consistent lighting and perspective.
• Character and visual identity consistency across a series of images, which is one of Nano Banana’s strongest aspects.

The site states that generation takes between ten and thirty seconds. The result is returned as a task that the page tracks until completion, and credits for failed tasks are automatically refunded.

Your images are public unless you disable this yourself

This is the most important thing to know before your first try. The “Public” option in the generator is enabled by default, and the page states that public images “will be displayed in the gallery.” The API is the same, as the default value for the isPublic parameter is true. If you are designing product images before launch, working for a client, or working on anything confidential, disable this option every time, or pass isPublic=false in every request.

As for data, both the terms and the privacy policy confirm that NanoCloud does not use your prompts, uploaded files, or results to train models, and the terms grant you full commercial rights to what you generate.

Tool for Removing the Gemini Watermark

Among the website’s tools is one called “Gemini Watermark Remover,” which runs in the browser and removes the visible mark from Gemini images using a technique that reverses transparent blending. The homepage also lists “watermark removal” among its editing features. Removing the mark from your own images may be legitimate, but this mark exists primarily to indicate that the image was generated by artificial intelligence, and hiding it may violate the rules of platforms that require disclosure of this. Indeed, NanoCloud’s own terms prohibit “misleading deepfakes or misleading content without clear disclosure.” Our view: do not use it to present generated images as real ones, and do not remove a watermark from an image you do not own.

Pricing: Points That Expire

Each generation consumes points. At the time of review, the documentation specified 5 points for nano-banana-fast, 10 for nano-banana, 20 for nano-banana-pro, and 20, 30, or 50 for nano-banana-2 depending on the resolution (1K, 2K, or 4K), with the highest VIP versions reaching 50 points at 4K resolution. The website generator, meanwhile, displayed 20 points per run in its default setting. The pricing page presents three purchase options, all alongside a struck-through “original” price:

• One-time purchase packages: about $9.90 for 2,900 points valid for one month, $19.90 for 7,000 points valid for three months, and $99.90 for 42,000 points valid for six months.
• Monthly subscriptions: about $9.90 for 4,250 points per month, $19.90 for 10,000 points with access to the API and professional models, and $99.90 for 60,000 points with unlimited API calls.
• Annual subscriptions are also listed, but the page data did not clarify whether the displayed price is monthly or annual, so verify this on the checkout page.

There are two notes. First, package points expire: the cheapest package gives you one month to use roughly 145 Pro-quality images. Second, the free tier is limited: daily check-in adds 10 points, or half a Pro image, and the site does not specify how many trial points a new account receives. The refund policy allows a full refund for an entirely unused package within seven days, and a refund for the unused portion of a subscription within fourteen days.

Arabic: The model may handle it, but the site does not say so

At the time of review, the interface was available only in English and Chinese. The homepage boasts of text accuracy within images “in English, Chinese, Japanese, and more” without mentioning Arabic, and nothing on the site addresses right-to-left text within images. We did not generate test images, as doing so requires an account and credits, so we cannot judge the quality of Arabic lettering in a poster or package design. If that is your requirement, first try a few short Arabic headings with the cheapest model, and check letter joining and word order before purchasing a larger package. Writing the prompt itself in Arabic is likely possible, but NanoCloud does not promise that either.

Who It Suits, and Who Is Better Off Looking Elsewhere

NanoCloud suits designers, marketers, and e-commerce store owners who want Nano Banana’s editing capabilities and character consistency through a simple browser interface. It also suits developers who prefer a single key and a single credit balance over an account on Google Cloud. The API documentation is clear, and the automatic refund of credits when a task fails is a nice touch.

On the other hand, those already comfortable with Google’s apps or developer tools can find the same models there without the intermediary’s markup. Teams handling confidential work should also be careful to disable the default public setting. The site does not mention the name of the company that operates it; the terms refer only to “NanoCloud and its operators,” so organizations that require a known entity and a data processing agreement should clarify this first.