Google Unifies AI Strategy with Launch of Gemini 3
Google releases Gemini 3 as its most capable multimodal model, deploying it simultaneously across consumer and enterprise platforms. The release features a powerful Deep Think reasoning mode designed for complex agentic tasks.
Google introduces Gemini 3 as its new flagship family of large multimodal models, marking a significant shift in how the company deploys its artificial intelligence. Unlike previous releases that trickled out to a few products over time, Gemini 3 launches as a unified platform that immediately powers consumer tools like Search and the Gemini app, as well as enterprise services like Vertex AI and the Antigravity IDE.
The initial rollout centers on Gemini 3 Pro, which serves as the primary model for multimodal understanding and agentic coding tasks involving text, code, and rich media. Developers can interact with the model through a massive context window of over one million tokens that accepts text, images, video, audio, and PDFs, while the system delivers up to 65,536 output tokens with structured JSON support.
Alongside the core Pro model, Google unveils Deep Think as a higher-intensity reasoning mode tailored for the most demanding workloads and long-horizon planning. This specialized engine drives advanced benchmark performance and builds on Google's previous gold medal victories at the IMO and ICPC competitions, establishing Gemini 3 as a state-of-the-art contender in the competitive AI landscape.