Google has released a preview of Gemini 3.1 Pro, the latest version of its flagship large language model, signaling another leap forward in the intensifying AI race.

The company said the model will be made generally available in the near future.

Early evaluations suggest Gemini 3.1 Pro significantly outperforms its predecessor, Gemini 3, which was already considered one of the most capable AI systems when it launched late last year.

Independent benchmark results shared by Google show strong gains across complex reasoning and knowledge-based tasks.

One benchmark, known as Humanity’s Last Exam, ranked the new model well ahead of earlier versions.

Additional praise came from Brendan Foody, CEO of AI startup Mercor, whose APEX benchmarking system measures real-world professional task performance.

Foody said Gemini 3.1 Pro now leads the APEX-Agents leaderboard.

The model’s performance highlights rapid progress in agentic AI, where systems can reason across multiple steps and complete tasks with minimal human input.

Industry observers note that these capabilities are increasingly central to enterprise adoption.

The release comes as competition intensifies among AI leaders, with rivals such as OpenAI and Anthropic also rolling out increasingly powerful models aimed at advanced reasoning and autonomous workflows.

Source: TechCrunch