From samosas with Altman to space visit: Create it all with Gemini’s update

If you’ve been online recently, you might have caught a whiff of something strange from Google. Something called the "Nano Banana," started making the rounds. Then, Google's own CEO, Sundar Pichai, dropped a simple post on X: three banana emojis.

The tech world loves a good mystery, and Google was having fun laying the breadcrumbs for what they just officially announced: a massive leap forward for their AI image model, now officially known as Gemini 2.5 Flash Image.

But let's be honest, "Nano Banana" is way more fun, and it’s the name that’s sticking. It’s a direct answer to one of the most persistent, frustrating problems in AI photo editing: the slow erosion of you.

Finally, an AI That Remembers Your Face

Seriously, we’ve all been there. You upload a great photo from your friend’s wedding and ask the AI, ‘Put me on a beach in Goa’ or ‘Change my kurta into a cool sherwani.’ The result comes back, and it's… weird, “Thoda Ajeeb sa”. Your nose looks a bit different, the small til on your cheek is gone, and your friends are commenting, “Bro, is that you?”. It’s like the AI found your long-lost, slightly-off twin.

This is that classic "identity drift" problem, and honestly, it's been the most annoying thing about AI editors. But Google claims its Nano Banana is here to be the bridge. Its killer feature, the one everyone’s talking about, is its solid ability to preserve your likeness. You can throw yourself into the middle of a Holi party, try on a 90s Bollywood hairstyle, or even see how your dog would look on a trip to Ladakh.

Through it all, the model holds onto your face, your features, and your pet’s unique look. Well, we tested it, still the perfection is far from the reality, but compared to its previous version, the results are impressive, especially if the reference is clear and sharp. Well apart from this, here are some more to expect in this new version:

• The Ultimate Photo Mashup (Blending Photos): Ever wanted a picture that’s impossible to take? Like you chilling with your pet on a trip you took solo, or getting your entire friend group into one perfect shot, even though two people were missing? Now you can. Just upload different photos, and the AI will stitch them together so seamlessly that it looks like one single, perfectly-timed click. It's the ultimate jugaad for your photo gallery.
• Edit Like You're Chatting (Multi-turn Editing): Forget trying to write one huge, perfect prompt and just hoping for the best. This feature makes editing a conversation. You can say, “Change the wall colour to blue.” Then look at it and go, “Cool, now add a plant in that corner.” Then, “Okay, one more thing, add a poster there.” You build your perfect image step-by-step, giving you total control instead of a one-shot gamble.
• Become an Art Director (Style Mixing): This is where you can get really creative. Love the vibrant, chaotic colour palette of a Chandni Chowk market photo? You can "steal" that entire vibe and apply it to a boring selfie. Or take the bandhani print from a picture of a dupatta and digitally "paint" it onto your sneakers. It lets you mix and match the mood and texture of different images to create something totally new and aesthetic.
• Keeping It Real, Literally (Responsible Creation): With all the deepfake drama and fake news circulating on WhatsApp, this is super important. Google is being upfront by putting a clear watermark on every AI-made image. There’s also a hidden digital tag (called SynthID) that proves it was made by AI. It’s a simple, no-nonsense way of saying, “This is cool, but it’s art, not reality,” so everyone knows what’s genuine and what’s not.

So Why Does This Matter Now?

This isn’t just a small tweak. Around launch, Gemini 2.5 Flash Image shot to the Number 1 spot on LMArena, a popular community leaderboard that collects millions of votes on AI models. That’s not Google’s own pat on the back, it’s a wide base of users giving the update a seal of approval.

With this release, Google is raising the stakes and going head-to-head with creative AI leaders like Midjourney, DALL-E, and Adobe Firefly. Its biggest selling point isn’t just sharp images, but consistency and control, features that matter to both casual creators and professional designers.

And here’s the kicker: it’s not locked up in a research lab. The update is live on both the free and paid tiers of the Gemini app, and it’s also available for developers through the Gemini API, AI Studio, and Vertex AI. In other words, Google’s most advanced creative tool yet is being placed directly into the hands of everyday users and builders. That move suggests the next phase of AI creativity is about to get a whole lot more personal.