Google rolls out Gemini 3.1 Pro AI model for complex tasks: Details

Google said that its latest Gemini 3.1 Pro model brings stronger reasoning, improved coding skills and higher usage limits to select Gemini app and NotebookLM users

Google has started rolling out Gemini 3.1 Pro, the latest upgrade to its artificial intelligence (AI) model lineup. The update is making its way to the Gemini app and is also available inside NotebookLM for eligible users. According to the company, this version improves the core reasoning abilities of Gemini, aiming to handle more layered and demanding tasks. In practical terms, that could change how people rely on AI for studying, analysing information, coding, or even working on creative ideas.

Smarter responses for everyday use

Google describes Gemini 3.1 Pro as a model built for situations where a quick summary just doesn’t cut it. Instead of offering surface-level answers, it’s meant to unpack complex topics, organise scattered information and present it in a more structured way. For users, this may translate into clearer explanations of tricky subjects, more organised research notes and better synthesis of large chunks of data.

 

To back its claims, Google pointed to improvements in benchmark tests. On ARC-AGI-2 — which measures how well an AI system can tackle unfamiliar logic problems — Gemini 3.1 Pro scored 77.1 per cent. The company says that’s more than double the reasoning performance of Gemini 3 Pro on the same test. While benchmark numbers don’t always reflect day-to-day use directly, they suggest the model should be more reliable when dealing with complicated queries.

Improvements in coding and creative work

One of the more practical additions is in code generation. Google says the model can create animated SVG graphics from a simple text prompt. Because these visuals are written in code rather than exported as video files, they stay sharp at any size and typically take up less storage. For developers, designers or even students experimenting with web projects, this could simplify certain tasks.

 

More broadly, Google positions Gemini 3.1 Pro as better at handling multi-step reasoning. That’s particularly relevant for tasks that involve planning, analysing multiple inputs or producing structured outputs instead of single-line answers.

Who gets access

Access to Gemini 3.1 Pro depends on the plan you’re on. In the Gemini app, the upgraded model is rolling out with higher usage limits for subscribers on Google AI Pro and Ultra plans. NotebookLM access is also restricted to Pro and Ultra users for now.

 

Developers can try the model in preview through the Gemini API via Google AI Studio, Gemini CLI, Google Antigravity and Android Studio. Enterprise customers, meanwhile, can access it through Vertex AI and Gemini Enterprise.

What happens next

Google calls this a preview phase, saying it wants to test and refine the improvements before a wider release. The company also mentioned that it is continuing to work on more advanced “agentic” capabilities, where AI systems can handle multi-step tasks more independently.