Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
The latest version of Google’s AI photo model is now available, thanks to… Gemini 3. Whether you choose to call it by its official name (Gemini 3 Pro Image) or by its well-known moniker (nanobanana pro), this model uses Gemini’s “deep thinking” feature to create better content.
Nanobanana pro promises to be even more capable, especially in the area of creating readable text, using up to 14 reference images and creating images at up to 4K resolution.
You can upload up to 14 separate reference images to nanobanana pro. Notice how each of the letters in the vector (left) appears unchanged in the final image (right).
Nanobanana pro is available now, rolling out globally in the Gemini app. In Gemini, select “Create Images” (with the 🍌 emoji) and switch your form to the Thinking form (located on the right side of the prompt window).
You can use nanobanana pro for free, but you’ll reach the generation cap faster than paying Google AI subscribers. These plans start at $20 per month. You will automatically be returned to the original form once your credits are exhausted. The model is also available in Gemini API, Google AI Studio, and… flow.
Inside the new banana nano model
AI-powered image generators are known to struggle to generate legible text; The distorted words were clear signs that the image had been created using artificial intelligence. OpenAI’s GPT-Image-1 was one of the first models to make significant progress in fixing this problem, however CNET test He showed that he was unable to create reliably readable text.
Google says the nanobanana pro is supposed to be better at this — and it’s so good that you can use the new model to create graphs, relying on Gemini’s integration with Google Search (and putting AI into search for paying Pro users) to help in the process.
Example of a chart prepared using nanobanana pro.
You may have noticed that none of the photos created with nanobanana pro have the diamond-shaped Google watermark, which usually indicates they were taken using AI. That’s because Google is offering a new way for its highest-paid subscribers (those who pay $250 a month for… Super plan) to remove those watermarks. Although it is limited to that top-level plan, it does exist It’s very difficult to know If something is created using artificial intelligence. Reducing watermarks will not help. Google says Gemini will be able to recognize any photos taken using its AI due to SynthID, an invisible watermark associated with AI-generated content. So, in theory, you could upload a photo to Gemini and ask if it’s real or made using Google’s AI. We’ll have to wait and see how well it actually works.
Don’t miss any of our unbiased technical content and lab reviews. Add CNET As Google’s preferred source.
The original banana nano model, the Gemini 2.5 Flash Image, was released this summer. Early testers thought the abilities were bananas. Throw in a series of fruit-themed teasers from Google executives, and the model quickly became known by its nickname nanobanana. Its popularity rose In the following days.
One of the things fans loved about the original Nano Banana was how consistent the character was kept. This means that nanobanan was able to edit your existing photos without distorting the characters or people in them. CNET’s hands-on testing found this to be true, but it’s still there Obvious errors in artificial intelligence. The new model may eliminate some of these errors.
It’s been a big week for Google, as it dropped a major update to its AI system, Gemini 3. The new models are rolling out now, aiming to give Gemini users a smarter system, Less flattering expertise.
For more, check out How to write the best AI guided image and Best AI image generators.