DeepMind Blog
Sep 30
3. 1 Pro is designed for tasks where a simple answer isn’t enough.
Our most specialized reasoning mode is now updated to solve modern science, research and engineering challenges.
Gemini 3. 5 is built to help you execute complex, agentic workflows.
The article "Welcome to Google Gemini 4 Argon" discusses how the new million‑token output window allows users to provide larger inputs to the model in a single pass. It examines benchmark results and cost implications to determine which users will benefit most from this capability.
Gemini 2. 5 Pro continues to be loved by developers as the best model for coding, and 2.