It’s only been a week since Google released its latest and most powerful AI model, Gemini 3, and it’s already become the AI ​​model of choice for many. Google is calling Gemini 3 a new era of intelligence, with the model reportedly beating stronger rivals like OpenAI’s ChatGPT in various benchmarks, including LM Arena. And now this model is a favorite of Salesforce CEO Marc Benioff. According to Benioff he was so impressed with Gemini 3 that he “is never going back to ChatGPT.”

Benioff, who says he has used ChatGPT every day for three years, shared his feedback in a post on X, praising Gemini 3 as a dramatic leap forward in logic, speed, and multimodal performance. He described the experience as transformative and claimed that the model “outperforms ChatGPT in logic, images and video” and delivers faster, faster responses across all formats. “I’ve used ChatGPT every day for 3 years. Just spent 2 hours on a Gemini 3. I’m not going back. The leap is crazy – logic, speed, images, video everything is faster and faster. It’s like the world has changed again.”

His comments followed a hands-on session with Google’s flagship model, which merges advanced text, image and video processing into a single system. The post immediately received millions of views and sparked lively discussion among developers, AI researchers, and enterprise users. While many echoed Benioff’s enthusiasm, others pushed back, questioning whether a single model could really make the leap in a field that evolves so rapidly. “The hype cycle is tiring. Every model trains on the same benchmark and incrementally improves every six months, doesn’t this become a moot point?” One user replied.

But Benioff isn’t the only one impressed by Gemini 3’s capabilities. Even OpenAI’s own CEO Sam Altman is not denying its power. Commenting on the launch, Altman commented that the Gemini 3 “looks like a great model”, while former Tesla AI director Andrzej Karpathy said he had a “positive initial impression”, calling it “very solid daily-driver capability” and placing it firmly in “tier-1 LLM” territory.

Initial reactions appear to support Google’s claims. CEO Sundar Pichai described Gemini 3 as the company’s “most intelligent model”, designed to integrate all previous Gemini capabilities so users can “bring any idea to life.” Google says that the main focus of this new AI model is multimodality, which is the ability to easily change between text, images, audio and video.

Gemini 3 is said to excel in logic and native multimodal understanding, analyzing text, video and files simultaneously to deliver superior accuracy and rich visual output. According to Google, its industry-leading one-million-token context window is a major technological advantage, allowing it to process large documents, large codebases, or multi-hour transcripts at once. The model reportedly shows significant advances in advanced tool use and long-horizon planning, enabling more sophisticated agentic workflows for enterprise tasks such as financial modeling, supply-chain optimization, and complex coding.