Google has introduced Gemini 3.1 Pro in preview, marking a significant advancement in its AI reasoning capabilities. The model achieves an ARC‑AGI‑2 score of 77.1%, more than double that of Gemini 3 Pro, and records an 80.6% score on SWE‑Bench Verified, a benchmark for autonomous software engineering tasks. These results underscore Google’s focus on enhancing AI performance in complex, real‑world applications. Gemini 3.1 Pro builds on the earlier Gemini 3 Deep Think upgrade and is positioned to support demanding tasks in science, research, and engineering. The preview release signals Google’s intent to push the boundaries of AI reasoning and practical utility. (blog.google)