Google Releases Gemini 3

Google released Gemini 3 on November 18, 2025 and made it available immediately in the Gemini app, developer platforms, and Google Search’s AI Mode.

The model emphasized stronger multimodal reasoning and agentic behavior. Google also introduced Deep Think for harder reasoning tasks and integrated Gemini 3 into its Antigravity development environment, where models could work across coding and tool-based workflows.

Launching a frontier model directly inside Search was especially significant. Search systems traditionally retrieve and rank existing documents; a generative model can instead synthesize an answer, create interactive interfaces, and decide which tools to use. That increases capability but also raises the cost of factual verification because generated output is not simply a ranked copy of one indexed page.