Gemini 3.1 Pro Preview is Google's latest preview flagship, positioned as an upgrade to the Gemini 3 Pro lineage with what Google describes as upgraded core intelligence. It is distributed widely through consumer and enterprise channels, reaching developers via the Gemini API in AI Studio, Vertex AI, and CLI tools, while consumer access comes through the Gemini app (with elevated limits for Google AI Pro and Ultra subscribers) and NotebookLM for Pro and Ultra users. The model carries a January 2025 knowledge cutoff and is offered in preview status, letting teams evaluate it before a general rollout.

Google's published evals highlight notable gains in reasoning-heavy and software-engineering tasks: a verified score of 77.1% on ARC-AGI-2 (more than double Gemini 3 Pro's prior mark), 94.3% on GPQA Diamond, and 80.6% on SWE-Bench Verified, as listed in the Gemini 3.1 Pro model card for February 2026. These results suggest practical strengths in novel pattern reasoning, graduate-level question answering, and real-world code generation, making the preview a strong fit for coding assistants, research workflows, and agentic applications that demand deep logical reasoning before the model moves to general availability.