Google announces Gemini 3 as battle with OpenAI intensifies

Google The search giant is launching its latest AI model, Gemini 3, as it races to keep up with ChatGPT creator OpenAI.

The new AI model will allow users to get better answers to more complex questions, “so you can get what you need with less prompting,” Alphabet CEO Sundar Pichai said in one of several blog posts Google published Tuesday.

Gemini 3 will be integrated into the Gemini app, Google’s AI search products AI Mode and AI Overviews, as well as its enterprise products. The rollout will begin Tuesday for select subscribers and roll out more broadly in the coming weeks.

The announcement comes nearly eight months after Google introduced Gemini 2.5 and 11 months after Gemini 2.0. OpenAI, which kicked off the generative AI boom with the public launch of ChatGPT in late 2022, introduced GPT-5 in August.

“It’s amazing to think that in just two years Al has evolved from simply reading text and images to reading the room,” Pichai wrote in one of his posts on Tuesday. “Starting today, we are launching Gemini at Google scale.”

The Gemini app currently has 650 million monthly active users and AI Overview has 2 billion monthly users, the company said. OpenAI said in August that ChatGPT reached 700 million weekly users.

Pichai added that the newest model is “designed to capture depth and nuance,” and said Gemini 3 is also “much better at understanding the context and intent behind your request, so you can get what you need with less incentive.” Google’s other AI models can still be used for simpler tasks, the company said.

Alphabet and its mega-capital rivals are spending heavily to build infrastructure for AI development and rapidly deliver more services to consumers and businesses. In earnings reports last month, Alphabet, Meta, Microsoft and Amazon lifted their guidance on capital spending, and they collectively expect that figure to rise to more than $380 billion this year.

Google said the AI ​​responses powered by Gemini 3 will “dispense with cliché and flattery for real insight – telling you what you need to hear, not what you want to hear,” according to a statement from Demis Hassabis, CEO of Google’s AI unit DeepMind. So are today’s AI chatbots, industry critics say flatterer.

Last week OpenAI released two updates to GPT-5. The company said one was “friendlier, smarter, and better at following your instructions,” while the other was “faster at simple tasks and more persistent at complex tasks.”

Google also announced a new agent platform called “Google Antigravity” that allows developers to write code at a “higher, task-oriented level.”

Gemini 3 is the company’s “best dither encoding model ever,” Josh Woodward, vice president of Google Labs and Gemini, told reporters at a briefing. Vibe coding refers to a rapidly growing market of tools that allow software developers to create code with prompts.

Google said the new model will enable “generative interfaces” and deliver specific answers in a way that resembles a digital magazine. As an example, the company asked Gemini to “explain the Van Gogh Gallery through the life context of each piece.” The result is a colorful, image-based description for each painting.

Within the scope of AI Mode, Gemini 3 will be available to paid subscribers first. Will be able to analyze a question and create a layout with visual elements such as images, tables, and grids. Google said it could make a custom-built interactive loan calculator or an interactive simulation of a complex physics problem.

Developers will be able to access the Gemini API and businesses will be able to integrate it through Vertex AI, the Google cloud service designed to build, deploy and manage AI models.

For enterprise customers, Gemini 3 can do things like create employee onboarding and training, more accurately analyze videos and factory floor footage, and manage purchasing, the company said.

WRISTWATCH: Google adds Gemini to Chrome for all users