By Jay Bonggolto
Publication Date: 2026-02-21 15:28:00
What you need to know
- Google is previewing Gemini 3.1 Pro, with a focus on complex, multi-step problem-solving that stumps most chatbots.
- The model scored 77.1% on the ARC-AGI-2 benchmark — more than double the reasoning performance of Gemini 3.0 Pro.
- You can use it across the Gemini app, NotebookLM, Vertex AI, and the Gemini API.
A few months after launching Gemini 3.0, Google is now previewing Gemini 3.1 Pro. This new model is designed to tackle complex, multi-step problems that most chatbots still find challenging.
This upgrade comes soon after Google launched its Deep Think features, which expanded the platform’s use in research, engineering, and science. Gemini 3.1 Pro is the intelligence behind these advances, and Google is now bringing it to consumer apps, developer tools, and enterprise platforms all at once.
You can already use Gemini 3.1 Pro in the Gemini app, NotebookLM, Vertex AI, and the Gemini API. Google wants this model available wherever people work and create.
