Google has officially launched Gemini 3, its cutting-edge foundation model, now accessible via the Gemini app and AI search interface. This release, just seven months after the introduction of Gemini 2.5, positions Gemini 3 as Google’s most powerful large language model (LLM) to date, marking a significant competitor in the rapidly evolving AI landscape. This rollout comes closely on the heels of OpenAI’s GPT-5.1 and Anthropic’s Sonnet 4.5, underscoring the relentless pace of advancements in AI technology.
In addition to the standard model, Google plans to introduce a more in-depth variant, Gemini 3 Deepthink, exclusively for Google AI Ultra subscribers in the upcoming weeks, pending successful safety evaluations.
Tulsee Doshi, Google’s head of product for the Gemini model, highlighted the model’s enhanced reasoning capabilities: “With Gemini 3, we’re seeing this massive jump in reasoning. It’s responding with a level of depth and nuance that we haven’t seen before.”
Independent benchmarks affirm these claims, with Gemini 3 achieving a groundbreaking score of 37.4 on the Humanity’s Last Exam, surpassing the previous record of 31.64 set by GPT-5 Pro. Additionally, Gemini 3 leads in user satisfaction tests, outperforming competitors on the LMArena benchmark.
As of now, the Gemini app boasts over 650 million monthly active users, and approximately 13 million software developers have integrated the model into their workflows.
Furthermore, Google has unveiled a new coding feature called Google Antigravity, which offers a multi-pane interface designed for agentic coding, akin to IDEs like Warp and Cursor 2.0. Antigravity merges a ChatGPT-style prompt interface with a command line and browser window, enabling users to see immediate effects of changes made by the coding agent.
“The agent can work with your editor, across your terminal, across your browser to ensure optimal application development,” stated DeepMind CTO Koray Kavukcuoglu.
This latest development from Google reiterates the company’s commitment to pushing the boundaries of AI while providing innovative tools for developers and users alike.
