[Google Releases Gemini 3.1 Pro, Inference Performance More Than Doubled Compared to Gemini 3 Pro] BlockBeats News, February 20: Google has released Gemini 3.1 Pro. Building on the Gemini 3 series, 3.1 Pro achieves a breakthrough in core inference capabilities. 3.1 Pro is a smarter and more powerful benchmark model for solving complex problems, showing significant progress in rigorous benchmark tests. In the ARC-AGI-2 benchmark test, which evaluates the model's ability to solve new logical patterns, 3.1 Pro achieved a validation score of 77.1%, more than doubling the inference performance of Gemini 3 Pro.