Google has released Gemini 3.1 Pro, an updated version of its flagship AI model. This new iteration is designed to excel at complex problem-solving and reasoning.
Gemini 3.1 Pro showed marked improvement in benchmarks. In Humanity's Last Exam, it achieved a score of 44.4 percent, surpassing its predecessor Gemini 3 Pro (37.5 percent) and OpenAI's GPT 5.2 (34.5 percent).
The model also demonstrated significant gains in ARC-AGI-2, a test for novel logic problems. Gemini 3.1 Pro reached 77.1 percent, more than doubling Google's previous score on this evaluation.
While Gemini 3.1 Pro leads in certain AI benchmarks, it is currently edged out by models like Claude Opus 4.6 in text generation and several models in coding tasks on platforms like the Arena leaderboard.