After seeing all the hype surrounding Google’s latest generative AI tool, I knew I had to try it out. The model is called Gemini 2.5 Flash Image. Updates Your ability to edit your photos in Gemini. AI enthusiasts call it the “nano-banana” model, driven by a series of banana-themed breakthroughs. Googling management.
In the few weeks since its launch, people have created more than 200 million AI images and more than 10 million people have signed up to use the Gemini app. After to Josh Woodward, VP of Google Labs and Gemini at Google.
Google has invested heavily in its generative media models this year, introducing updated versions of its image and video generator models at its annual I/O developer conference. Google’s AI video generator, Veo 3, surprises with synchronized sound, a first among the AI giants. And YouTubers have created more than 100 million AI videos using Flow, Google’s AI video tool.
I’ve spent a lot of time testing AI design software and was excited to see what Google came up with. But my testing of Flash Image 2.5 has shown that just because something has a prominent mention doesn’t mean it will always meet your expectations. This was my experience with the Gemini Nano Banana: the good, the bad and the frustrating.
What worked?
The Gemini Banana template is incredibly good at adding elements to existing images and integrates AI-generated elements well into any image you create. Also, character consistency remains fairly stable, meaning the people in my photos weren’t too distorted or shaky after the AI processing. Both are important distinctions for AI imaging programs and something Google says it’s working to improve.
You can see both features in this photo of me and my sister. Our overall look remained unchanged in the edited version (right), showing the consistency of the characters. I asked Gemini to add a third sister who looked like the two of us, which he surprisingly did and added a third wife between the two of us.
I was also very impressed with how quickly Gemini was able to produce finished images. Anything under a minute earns me a gold star, and Gemini routinely processes requests in less than 15 seconds. I also appreciate that you added a watermark to all the images you created and edited. While I don’t like how tech companies have corrupted shiny emojis for AI, having labels for AI-generated content is extremely important. Google’s SynthID and behind-the-scenes work also help distinguish AI content from human-generated images.
Gemini is also good for creating full AI images, but I recommend using Image 4 or another AI imager – these have more convenient controls and settings that allow you to get closer to what you want with less effort.
Which really didn’t work
Gemini bananas have serious limitations. It automatically generates square images and follow-up requests to scale images to other sizes were ignored or failed.
I also noticed that Gemini lowered the resolution of many of my images. I mainly shoot with my iPhone 16 which has excellent cameras, but after testing the Gemini banana model these small details were often blurry. It’s boring and won’t convince any photographer.
I have tried several times to ask Gemini to make changes to pictures that I would have had difficulty doing manually. This is one area of photo editing where AI should excel: automating trivial but detail-intensive edits. Unfortunately, the Gemini had trouble following quickly, so it didn’t do what I asked.
I tried several times to get the Twins to remove the reflections from a cutout of the Freakier Friday movie poster, but they were still there. And the more I tried to remove reflections, the worse the image quality got with each request. The once-clear text became illegible after I finally gave up, not to mention the odd accidental damage to Lindsey Lohan and Jamie Lee Curtis’ faces.
The Gemini nanobanana had difficulty creating images of different sizes. Resizing and cropping images is an important photo editing process, but Gemini didn’t or couldn’t include simple resizing guidelines in my instructions.
I reached out to Google about the fix and sizing issues, and a spokesperson said the tech company was “aware of both issues and actively working on them. This was a significant update from our previous model, but we will continue to improve the model.”
Overall, the “Gemini Nano Banana” showed me that Google is serious about maintaining its dominance in the generative media space. But it comes with significant drawbacks as too much emphasis is placed on generating new assets rather than using AI to improve and solve common photography problems. For now, the Nano Banana model is best suited for Gemini fans who want to make big changes quickly. Those of us looking for more accurate tools will have to wait for the next big Google update or look for another program.
Gemini Nano Banana Availability, Price and privacy Policy
You don’t need to do anything to access the new template; It is automatically added to the Gemini 2.5 Flash base model. Gemini is available for free. Additional templates and higher usage limits are available in Google AI plans from now on $20 per month.
If you are a paid subscriber, you may also be able to access the model through Google AI Studio. From there, all you have to do is upload an image and enter your message. Each message uses between one thousand and two thousand tokens, depending on the level of detail required. Adobe Express and Firefly users can now also access the new template.
Googling Gemelli’s privacy policy claims it can use the uploaded information to improve its AI products, so the company recommends avoiding uploading sensitive or private information. The company Prohibited Use of AI Policy It also prohibits creating illegal or offensive material.
To learn more, check out the best AI image generators and everything announced at the Made by Google Pixel 10 event.