On February 19, 2026, Google introduced Gemini 3.1 Pro in preview, marking a significant advancement in AI reasoning capabilities. The model achieved an ARC‑AGI‑2 score of 77.1 percent—more than double that of Gemini 3 Pro—and scored 80.6 percent on SWE‑Bench Verified, a benchmark for autonomous software engineering tasks. (en.wikipedia.org)
This release builds on Google’s earlier upgrades to Gemini 3 Deep Think, which targeted practical applications in science, research, and engineering. The new model underscores Google’s continued investment in AI systems capable of complex, multi‑step reasoning. (en.wikipedia.org)
Gemini 3.1 Pro’s performance leap reflects a broader trend in AI development: moving beyond generative capabilities toward models that can reason, plan, and execute tasks with greater autonomy. The improved scores on ARC‑AGI‑2 and SWE‑Bench Verified suggest that Gemini 3.1 Pro is better suited for high‑stakes domains such as scientific research and software engineering.
By releasing Gemini 3.1 Pro in preview, Google is signaling its intent to push the boundaries of AI reasoning. The model’s enhanced benchmarks may influence enterprise adoption and shape expectations for future AI systems capable of tackling complex, real‑world problems.