Google unveils updated AI model Gemini 3.1 Pro
Google has released an updated version of its flagship artificial intelligence model, Gemini 3.1 Pro. The announcement highlighted the ARC-AGI-2 test, which evaluates the model’s ability to solve new logical patterns under uncertain conditions.
The model scored 77.1%, more than double the previous Gemini 3 Pro version’s result. Google states that the new version has made significant progress in core reasoning skills and improved performance on complex problems. It also shows competitive results in coding and working with AI agents.
Gemini 3.1 Pro represents a major step forward in AI development, opening new opportunities for the technology’s application across various industries.
