Gemini 3.1 Pro Preview is positioned as Google's most advanced reasoning model in the Gemini family, designed for complex problem solving across diverse inputs and agent-driven workflows. Official Google Cloud documentation on Vertex AI describes it as the flagship reasoning model in the lineup, while the OpenRouter provider listing emphasizes its frontier status with enhanced software engineering performance, improved agentic reliability, and more efficient token usage. It belongs to the gemini-pro family, sitting at the top of Google's reasoning tier as a preview release intended for developers building demanding applications. The model is engineered to handle long-context reasoning, multimodal comprehension, and tool-heavy agent workflows, making it suitable for tasks that require deep analysis rather than lightweight generation.

The model supports multimodal inputs including text, audio, images, video, PDFs, and code, producing text output through its API. It operates within a that quick-info value window of the cataloged API limit tokens with a maximum output of the cataloged API limit tokens, giving it substantial room for long documents, codebases, or multi-turn agent interactions. For workloads that mix a general bash shell with registered third-party or user-defined functions, a companion variant called Gemini 3.1 Pro Preview Custom Tools exists to improve tool selection reliability and prevent overuse of the bash tool. Pricing is tiered, starting at $the listed price per million input tokens and $the listed price per million output tokens for prompts up to the listed price tokens, rising to $the listed price and $the listed price respectively above that threshold, with prompt caching available as a cost lever for repeated context. Vertex AI hosts the model with a documented Gemini 3.1 Pro page, making it directly accessible to enterprises integrating through Google Cloud.