New Delhi: Tech billionaires keep weighing in on which AI model is winning, and Marc Benioff is the latest to pick a side. This week, the Salesforce founder and Time magazine owner declared that Google’s newly launched Gemini 3 is far better than ChatGPT and that he’s a permanent convert. His post on X has kicked off a new Gemini vs ChatGPT debate in the AI world.
Benioff wrote he used ChatGPT for three years, Gemini 3 for two hours, and he isn’t going back.
“Holy shit. I’ve used ChatGPT every day for 3 years. Just spent 2 hours on Gemini 3. I’m not going back. The leap is insane — reasoning, speed, images, video… everything is sharper and faster. It feels like the world just changed, again,” Benioff posted, three days after Google unveiled Gemini 3, its most powerful AI model to date.
Holy shit. I’ve used ChatGPT every day for 3 years. Just spent 2 hours on Gemini 3. I’m not going back. The leap is insane — reasoning, speed, images, video… everything is sharper and faster. It feels like the world just changed, again. ❤️ 🤖 https://t.co/HruXhc16Mq
— Marc Benioff (@Benioff) November 23, 2025
A flurry of comments followed. Some asked, “What took you so long?” Others predicted: “Give it few weeks… you will come back to Grok and ChatGPT.”
So, what really is the difference between Gemini 3 and ChatGPT 5.1?
Also Read: What is ‘Posting Zero’? The internet generation is growing tired of social media
What are the new features of Gemini 3?
Google CEO Sundar Pichai outlined five new features of Gemini 3 in a thread on X. He said the model can understand many types of inputs, including photos, PDFs, rough sketches, and diagrams.
1/ Gemini 3 in action, a quick🧵
You can give Gemini 3 anything (images, pdfs, scribbles, etc) and it will create whatever you like: an image becomes a board game, a napkin sketch transformed into a full website, a diagram could turn into an interactive lesson. pic.twitter.com/xCry35mFbW
— Sundar Pichai (@sundarpichai) November 18, 2025
The new model can also watch long videos and break them down. Pichai said it can analyse sports clips, point out errors, and even suggest drills for improvement, due to its stronger visual and spatial reasoning.
According to the Google CEO, Gemini 3 is changing the way Search works. Instead of only producing written answers, the model can create visual layouts and interactive tools. Additionally, it shows photos, modules, and interactive sections that can be tapped into.
Then there’s Gemini Agent, which Pichai said, “uses Gemini 3’s advanced reasoning to break down complex tasks, like booking local services or organizing your inbox, into actionable steps.”
Alongside Gemini 3, Google introduced Nano Banana Pro, an image-generation and editing model that promises studio-grade visuals, reliable text rendering, multilingual support, consistent brand styling, and high-resolution output including 4K.
Gemini 3 is one of Google’s most aggressive bids to reclaim AI leadership as OpenAI rolls out the GPT-5 series. Benioff’s public endorsement adds momentum, especially with him directly pitting the two LLMs against each other.
Gemini 3 vs GPT-5.1 on pricing and rankings
The release of Gemini 3 comes close on the heels of ChatGPT 5.1, with both promising heightened capabilities. But what are their pricing and performance differences?
Google calls Gemini 3 its “most intelligent model yet,” while OpenAI has labelled GPT-5.1 as “smarter” and “more conversational.”
But there is a clear difference in pricing for application programming interface (API) users — the companies and developers who plug these models into their own apps — with ChatGPT being easier on the pocket.
GPT-5.1 costs $1.25 (Rs 111) per million input tokens and $10.00 (Rs 892) per million output tokens. A ‘token’ is a small chunk of text, and models bill users based on how many tokens they send to the API.
Meanwhile, Gemini 3 Pro lists token-based tiers at $2.00 (Rs 178) per million input tokens and $12.00 (Rs 1,070) per million output tokens. It supports contexts up to approximately 200,000 tokens, which essentially means it can handle very long inputs in one go.
For regular users, both Gemini 3 and GPT-5.1 offer paid plans starting at about $20 (approximately Rs 1,800) a month, with higher-end tiers priced above that.
On performance, Gemini 3 is currently sitting at the top of the LMArena leaderboard, a public ranking that tests models on reasoning, coding, writing, and factual accuracy. It scores 1495, followed by two Grok 4.1 variants at 1481 and 1462. GPT-5.1 comes in at number four with 1454, just three points ahead of Gemini 2.5 pro.
Also Read: Deepfake on duty: when I asked AI to read Op Sindoor citations
When it comes to handling many types of data and very long inputs, Gemini 3 is the stronger model. Google says it can take in large amounts of text at once, move easily between images, text, and code, and use a “Deep Think” mode to plan and reason across complex tasks.
GPT-5.1, meanwhile, is designed mainly for high-quality conversation. It’s strong at writing, following instructions, and adjusting tone or personality to suit what the user wants. The paid tier also has a memory function across chats.
Tom’s Guide, a popular tech how-to and product-testing website, ran multiple tests on both models and found that GPT-5.1 doesn’t match Gemini 3’s breadth in modality or long-form reasoning.
“Gemini 3 emerged as the clear winner, taking seven out of eleven rounds with consistently superior performance in creative constraint-following, UX design thinking, critical analysis, strategic reasoning and cross-domain integration,” read the site’s verdict.
However, it did concede a few points to GPT-5.1, noting that “it excelled in mathematical reasoning and coding logic”.
For everyday users, however, GPT-5.1 feels more familiar and easy to use; an informal run-through of Reddit discussions shows that its memory feature and general UX are still big pluses compared to Gemini 3.
(Edited by Asavari Singh)