Google has officially released Gemini 3.1 Pro, a substantial upgrade to its Gemini 3 Deep Think model, aimed at enhancing reasoning capabilities for complex tasks in science, research, and engineering. The model is now available across Google’s consumer and developer platforms, including the Gemini app, Google Search, AI Studio, and Vertex AI.
Gemini 3.1 Pro achieves an ARC‑AGI‑2 score of 77.1%, more than double that of Gemini 3 Pro, and scores 80.6% on the SWE‑Bench Verified benchmark, which evaluates autonomous software engineering tasks. These results mark a significant leap in core reasoning performance. The rollout began on February 19, 2026, following a major update to Gemini 3 Deep Think the previous week.(blog.google)
The upgrade underscores Google’s continued push into agentic AI—models capable of multi-step, autonomous reasoning. By integrating Gemini 3.1 Pro across its ecosystem, Google aims to bring advanced intelligence to everyday applications, from developer tools to search and enterprise services. The improved benchmarks suggest the model is better equipped to handle complex logic, coding, and scientific reasoning tasks.
This release positions Google strongly in the competitive AI landscape, where reasoning performance is increasingly a differentiator. With Gemini 3.1 Pro’s enhanced capabilities, Google may further challenge rivals in both consumer-facing and enterprise AI markets.
Tags: [“research”, “industry”]