These days a novelty has emerged for AI model enthusiasts, especially for users seeking an image generator capable of making the giants in the industry pale. Nano Banana is the phenomenon of the moment; it seems that behind this cute nickname hides one of the world’s most powerful tech giants. Will it really be so? In the meantime, let’s take stock of the situation with all the details known so far.
Nano Banana is a mysterious AI image generator that appeared practically from nowhere: it has captivated everyone with its power, but it emerged without pomp, press releases or anything else. The AI model was discovered inside the Battle section of LMArena: the open-source platform allows comparing and evaluating different large language models (LLMs) and the Battle mode allows you to test two of them blind.
Indeed, Nano Banana has surprised everyone thanks to its ability to produce accurate images, to interpret instructions at multiple levels and to edit photos with extraordinary quality. See for yourself: the post below presents various tests, both with real photos and with images created directly by the AI model (with results that even rival Photoshop).
The most widespread hypothesis is that Nano Banana is the new AI image generator developed by Google (and therefore directly connected to Imagen and Gemini), currently in testing in a secret capacity. As of this writing there is still no official confirmation, so for now the origin of the model should be taken with a grain of salt (and considered as a rumor).
Trattandosi di un generatore di immagini basta inserire un prompt dettagliato e dare il via all’operazione.
Currently Nano Banana is available randomly via LMArena Battle Mode or thanks to Flux-AI, in this case directly.
In addition to generating images via prompts, Nano Banana can also modify user-uploaded photos. It is possible to perform any modification without the need for complicated tools: simply write the prompt to improve sharpness, change the subject’s expression or position, and much more, all with a precise and realistic final effect (for real subjects).
Nano Banana’s performance is truly excellent both in editing and in understanding commands and in response times. In a few seconds you can obtain a high-quality image; it is also able to understand multi-step commands with precision, maintaining a good level of coherence. You can change subjects and environments, lighting and perspective, all in one go and without errors. All this while switching between styles without any problem.
As of this writing the AI model is available only through third-party platforms: you can use it by subscribing and purchasing credits, but it’s wise to wait for the official debut.
During use of the model there have been occasional defects such as anatomical anomalies (hands or fingers), issues with text elements inside images, or some inconsistencies in lighting when dealing with complex environments.
The launch of iPhone Duo, instead of pushing customers toward the new Apple device, seems…
iQOO confirmed with an official teaser that Pad Ultra will debut on September 29, presenting…
OPPO has confirmed the specifications of Watch S2 just days before the official debut, set…
OpenAI has announced the withdrawal of GPT-5.5 starting from October 14, across ChatGPT, Work and…
The wait for the arrival of the Samsung Galaxy S27 series is growing day by…
Between lectures, digital notes, online meetings and projects to carry out even outside the office,…