Google’s Gemini 3 AI Takes on ChatGPT: Revolutionary Model Unveiled!

On Tuesday, November 18, 2025, Google introduced Gemini 3, a new suite of language models touted as “the smartest” and intended to mark a significant step forward in the pursuit of general artificial intelligence, according to a press release. The preview version of Gemini 3 Pro is being rolled out “starting today” across several of the company’s products, including the namesake app, AI Studio, Vertex AI, and AI Mode, a conversational search interface not yet available in France.

Google Counters OpenAI with Gemini 3

Just days after the release of GPT-5.1 by OpenAI, which aimed to address some limitations of GPT-5 and make it “warmer,” Google has responded by unveiling a model described as its rival.

Described as the “best model in the world for multimodal understanding,” capable of ingesting and interpreting multiple types of data simultaneously, Gemini 3 Pro focuses on providing “intelligent, concise, and direct responses,” the company explains. Google, sprinkling its announcement with digs at its main competitor in the generative AI market, asserts that the successor to Gemini 2.5 moves away from “clichés and flattery towards genuine insight, telling you what you need to hear, not just what you want to hear,” aiming to be a real “thought partner.”

Gemini 3 is also far better at understanding the context and intent behind your query, so you get what you need with less prompting,” celebrates Sundar Pichai, CEO of Google.

Gemini 3 Pro, Already Leading the LMArena Rankings

While Google doesn’t shy away from using superlatives in its press release, and only real-world testing will confirm their validity, one thing is certain: from its preview release, Gemini 3 Pro has quickly topped the LMArena rankings, which measure the performance of models in real-time based on user votes. Google claims it also excels in numerous benchmarks related to academic reasoning, scientific knowledge, and mathematics.

A More Advanced Deep Think Version Still in Development

Alongside the Pro version, Gemini 3 comes in a second iteration named Deep Think, which pushes “the boundaries of intelligence even further” to “help you solve more complex problems,” Google states. Still in development and also excelling in various benchmarks, this version will undergo additional security evaluations and testing before being rolled out to subscribers of Google AI Ultra. The company did not specify whether this model series would include a Flash version, like Gemini 2.5, which is faster and less costly. Nonetheless, the press release notes, “We plan to launch more models from the Gemini 3 series soon to further enhance your AI capabilities.”

Gemini 3 Pro is available today for all users through the Gemini app. For developers, it is accessible via the Gemini API in AI Studio, within Google Antigravity, their new agentive development platform, and through the Gemini CLI. It has also been integrated into Gemini Enterprise.