Gemini 3.1 Pro represents an advanced iteration in core reasoning capabilities within Google's Gemini series, designed specifically for complex problem-solving scenarios where simplistic answers are insufficient. This model is presented as a more capable baseline intelligence, building upon its predecessor, Gemini 3 Deep Think.

The core methodological advancement, as demonstrated by its performance and capabilities, lies in significantly enhanced reasoning and synthesis. This is quantitatively supported by its achievement of a 77.1% verified score on the ARC-AGI-2 benchmark, a test evaluating a model's ability to discern and apply entirely novel logic patterns, indicating a more than twofold improvement over Gemini 3 Pro's reasoning performance.

The model's improved intelligence is applied across various sophisticated tasks, showcasing its ability to perform advanced reasoning through:

1. Code-based Generative Synthesis: The capacity to generate functional, production-ready code directly from natural language prompts. For instance, it can produce animated Scalable Vector Graphics (SVGs) that are resolution-independent and have minimal file sizes, implying a deep understanding of visual descriptions and their translation into vector code.
2. Complex System Integration and Configuration: Demonstrated by its ability to synthesize dynamic systems. An example provided is the creation of a live aerospace dashboard from a text prompt, successfully configuring public telemetry streams (e.g., for the International Space Station's orbit). This capability suggests robust reasoning for bridging high-level design requirements with technical API specifications and real-time data integration.
3. Interactive and Multimodal System Design: The model can generate intricate, interactive experiences, such as a 3D starling murmuration that responds to hand-tracking input and features a generative audio score that adapts to the birds' movement. This highlights its capacity for sophisticated 3D code generation, real-time input processing, and multimodal content synthesis (visual, auditory, interactive).
4. Creative Contextual Reasoning and Application: The model exhibits advanced semantic understanding and creative application by translating abstract themes into functional code. An example is designing a contemporary web portfolio for Emily Brontë's "Wuthering Heights," where it reasons about the novel's atmospheric tone to create a thematically appropriate user interface, rather than merely summarizing text. This demonstrates the model's ability to extract nuanced contextual meaning and manifest it through design and code.

These capabilities collectively suggest a methodological leap in inferential reasoning, cross-domain knowledge transfer, and the ability to decompose complex, multi-faceted problems into actionable, generative outputs, supporting ambitious agentic workflows.

Gemini 3.1 Pro is rolling out in preview across various platforms: for developers via the Gemini API (Google AI Studio, Gemini CLI, Google Antigravity, Android Studio), for enterprises in Vertex AI and Gemini Enterprise, and for consumers through the Gemini app and NotebookLM. Initial access includes higher limits for Google AI Pro and Ultra plan users.