Google announces Gemini 3.1 Pro, says it's better at complex problem-solving

Ars Technica

by Ryan Whitwam

February 19, 2026

AI-Generated Deep Dive Summary

Google has unveiled its latest AI innovation, Gemini 3.1 Pro, touting enhanced capabilities for complex problem-solving and reasoning. This upgraded version builds on the foundation of Gemini 3, which was released just months ago in November, and aims to deliver significant improvements in tackling intricate challenges. The announcement highlights that Gemini 3.1 Pro is now available in preview for developers and consumers, promising a more advanced tool for addressing demanding tasks. The new model has demonstrated notable progress in benchmarks, particularly in specialized knowledge domains. In the rigorous "Humanity's Last Exam," designed to test advanced domain-specific understanding, Gemini 3.1 Pro achieved an impressive score of 44.4 percent, outperforming its predecessor Gemini 3 Pro (37.5%) and OpenAI's GPT 5.2 (34.5%). This marks a significant leap in the model's ability to handle nuanced and highly technical queries. Additionally, Google has integrated Gemini 3.1 Pro into its Deep Think tool, further enhancing its problem-solving capabilities for users. For tech enthusiasts and professionals, this development underscores Google's commitment to pushing the boundaries of AI innovation. While the improvements may appear modest compared to previous updates, the enhanced reasoning and domain-specific knowledge make Gemini 3.1 Pro a valuable asset for enterprise applications, research, and customer support. As AI continues to evolve, such advancements will play a crucial role in shaping industries that rely on intelligent systems to address complex issues. The release of Gemini 3.1 Pro also reflects Google's competitive stance in the AI landscape, particularly as other tech giants like OpenAI refine their models. By focusing on specialized knowledge and problem-solving, Gemini aims to carve out a niche in sectors where tailored expertise is critical. As more businesses integrate AI into their operations, the capabilities of tools like

Verticals

techscience

Originally published on Ars Technica on 2/19/2026