Krux
February 21, 2026
Google Drops Gemini 3.1 Pro With Wild Reasoning Upgrade
Published: February 21, 2026 at 3:43 AM
Updated: February 21, 2026 at 3:43 AM
100-word summary
Google just launched Gemini 3.1 Pro, a significantly smarter AI model built for complex problems that need more than quick answers. It scored 77.1% on the ARC-AGI-2 benchmark, literally doubling the reasoning power of its predecessor. The upgrade rolls out today across the Gemini app, NotebookLM, API, and enterprise tools. What makes it wild? It can generate animated website code from text, build live aerospace dashboards tracking the ISS, and even code 3D bird simulations with hand-tracking. This positions Google to compete harder in agentic AI workflows where models actually solve multi-step problems autonomously, not just chat.
What happened
Google just launched Gemini 3.1 Pro, a significantly smarter AI model built for complex problems that need more than quick answers. It scored 77.1% on the ARC-AGI-2 benchmark, literally doubling the reasoning power of its predecessor. The upgrade rolls out today across the Gemini app, NotebookLM, API, and enterprise tools. What makes it wild? It can generate animated website code from text, build live aerospace dashboards tracking the ISS, and even code 3D bird simulations with hand-tracking.
Why it matters
This positions Google to compete harder in agentic AI workflows where models actually solve multi-step problems autonomously, not just chat.