Technology · 20 February 2026
Gemini 3.1 Pro arrives with double the reasoning power
Photo by Rubaitul Azad on Unsplash
Google introduced its newest artificial intelligence model on 19 February 2026, designed for difficult problem-solving. This update achieves a 77.1 per cent score on the ARC-AGI-2 benchmark, doubling the previous version's reasoning capabilities. Users can access the tool through Vertex AI and the Gemini app to process text, video, and code. It features a context window of one million tokens for handling large amounts of data.
Reported by Google Blog · How we write briefs