When ChatGPT broke onto the scene a couple of years in the past, Google’s strategy to AI was to place the know-how in as many locations as potential, hoping it might resonate with customers. Initially, it did not work — ChatGPT had higher fashions, extra use circumstances, and superior model energy.
Over the previous 12 months, there’s been a powerful sense that the tide is popping. Gemini is the one AI chatbot to constantly achieve internet visitors share over the previous 12 months, chipping away at ChatGPT’s main lead (by way of SimilarWeb on X). On prime of that, Gemini 3 is right here, and it has been so well-received by customers that Google is taking inventive steps to satisfy the excellent demand.
Final month, I wrote that Google’s Veo 3 mannequin could not maintain a candle to OpenAI’s Sora 2 as a result of the corporate did not perceive how a lot wrapping AI fashions in a user-friendly, brand-positive bundle mattered.
Since then, we have seen Gemini 3 and Nano Banana Professional debut, and Google has regained my belief. Google confirmed that it might probably constantly construct nice fashions, wrap them in partaking apps, and create catchy manufacturers for its AI fashions. In consequence, I am a believer.
Gemini 3 reveals Google can keep on the prime
If we have realized something over time, it is that constructing one nice AI mannequin is not sufficient for sustained success. OpenAI was as soon as the one one with a top quality massive language mannequin (LLM), and now its GPT-5.1 mannequin sits outdoors the highest 5 on the LMArena leaderboard. DeepSeek blew up in January 2025 and pale into obscurity once more simply as rapidly when different corporations caught up.
So, when Gemini 2.5 Professional dominated the LMArena and WebDev Area leaderboards, I nonetheless wanted to see extra. Certain sufficient, Gemini 2.5 Professional was finally unseated by fashions from xAI, Anthropic, and OpenAI. That is why Gemini 3 Professional’s return to the highest of the chart after Gemini 2.5 Professional was dethroned is an efficient signal of Google’s consistency.
Gemini 3 Professional stays the top-ranked textual content mannequin on the LMArena leaderboard. Google has the most effective mannequin within the textual content, imaginative and prescient, text-to-image, picture enhancing, search, text-to-video, and image-to-video classes. That is an impressive exhibiting — the solely class Google does not lead is WebDev Area. Gemini 3 Professional debuted within the prime spot, however was bumped right down to second place by Claude Opus 4.5.
Google’s physique of labor now conjures up a variety of confidence. It is not simply able to making industry-leading fashions. It is also able to constantly constructing new ones whereas going through threats from competing corporations.
Google is aware of there’s rather more to AI than fashions
One other promising ingredient of the Gemini 3 launch is the mannequin’s availability. It debuted instantly throughout the Gemini app, and AI Mode in Search, and Google’s developer instruments — together with Google AI Studio and Vertex AI. It is the primary time a brand new AI mannequin has been obtainable in Google Search, the corporate’s flagship software program product, on day one.
A mannequin is simply nearly as good as its availability and person interfaces, and Google is telling us it understands that with the Gemini 3 rollout. It made a number of new methods to work together with Gemini 3 alongside the brand new mannequin.
There’s the Gemini Agent, an agentic platform that may take actions in your behalf inside and past the Google ecosystem. AI Mode in Search added generative UIs, that are customized and interactive visuals created with Gemini 3 in actual time. Most significantly, Google launched a brand-new app with an agent-first built-in improvement surroundings (IDE) in Antigravity.
Lastly, Google clearly will get that AI branding issues
OpenAI confirmed simply how a lot branding issues within the AI race with Sora, the viral app for AI-generated movies. Seemingly in a single day, the complete notion of AI movies modified. What as soon as was referred to as AI “slop” all of a sudden turned enjoyable and interesting. Google should have paid consideration, as a result of Nano Banana follows the very same blueprint.
Nano Banana Professional is basically the Gemini 3 Professional Picture mannequin, however the latter title is not very enjoyable, is it? As a substitute, Google gave this text-to-image mannequin a unusual title and a banana emoji. I don’t know what Nano Banana means, or why there is a banana emoji plastered in every single place, together with in Google Messages.
Nevertheless, I do know what Nano Banana is not. It isn’t named like a scary AI image-generation mannequin. It isn’t branded like an enterprise or improvement device. Nano Banana creates AI pictures for the on a regular basis individual, identical to Sora did for movies.
Gemini already has a fairly robust model, as it is a fixture on Android and Put on OS, and shortly can have a stronger presence in Google Dwelling and Android Auto. Nevertheless, Google’s generative picture and video fashions simply fell into the “slop” class earlier than Nano Banana made picture fashions quirky and approachable.
Gemini 3 is a significant turning level for Google’s AI efforts
Gemini 3 could not have come at a greater time for Google. The corporate wanted to point out it understood what actually makes a terrific AI ecosystem, and it did simply that.
Gemini 3 Professional and Gemini 3 Professional Picture exhibit that Google can reply when challenged with new fashions from xAI, Anthropic, and OpenAI. New apps like Google Antigravity are proof that the corporate is aware of the way in which you work together with AI fashions is simply as essential because the fashions themselves. And crucially, Nano Banana is a powerful sign that the Mountain View corp. lastly grasps that model energy is social foreign money within the AI race.
Google’s AI efforts are firing on all cylinders, and I am unable to wait to see what else it builds.





