Building with Gemini 3.1 Pro
We’re releasing Gemini 3.1 Pro today.
3.1 Pro is a noticeably smarter, more capable baseline for complex problem-solving. We’re shipping 3.1 Pro at scale – including across Vertex AI and Gemini Enterprise – and building upon our goal to help you transform your business for the agentic future.
3.1 Pro is designed for tasks where a simple answer isn’t enough, taking advanced reasoning and making it useful for your hardest challenges. This can help in your practical applications — from bringing a creative project to life to building complex visualizations.
Here’s five ways Gemini 3.1 Pro changes the way you build:
1. A clear upgrade for complex tasks
We know reliability is a critical metric for developers. Gemini 3.1 Pro represents a step forward in core reasoning:
2. Taking advanced reasoning and making it useful
Building complex software requires proper planning. Gemini 3.1 Pro is designed to solve tougher problems, giving you the reasoning depth your business needs. You can map out a robust plan before the model generates an output, which makes it easier to navigate repositories, run tests, debug errors, and work through solutions.
Benchmark numbers are great, but they need to translate to actual utility. So we tested Gemini 3.1 Pro on tasks that require both logic and creativity.
3. One-shot full stack development
Gemini 3.1 Pro is great for full stack development, including generating beautiful front-end UIs that include animations and data. Here’s a Wuthering Heights demo:
4. Complex visual simulation & physics
The model also shines at complex visual simulation and physics. That includes being able to generate full Three.js scenes with many moving dynamic elements and shadows.
Recommended by LinkedIn
5. Creative coding
This model is also great for creative coding. Many of us enjoy looking at music visualizers, but you can actually have it generate complete shaders for you, such as in this demo:
Build with Gemini 3.1 Pro: RepotoViralVideo (Open-source tutorial)
To help developers explore these new possibilities, we're sharing an open-source app powered by Gemini 3.1 Pro.
RepotoViralVideo turns any GitHub repository into a promo video in just one click. We built it to demonstrate how Gemini 3.1 Pro handles complex, multi-step media generation.
Here's a sample video generated for awesome-llm-apps (95K+ ⭐):
Get the code and try it out (fork and star): https://github.com/Shubhamsaboo/repotovideo
Get your API from Vertex AI Studio or Google AI Studio.
This article is authored by Shubham Saboo and Addy Osmani . Reach out to them if you have any feedback or want to share cool things you are doing with Gemini 3.1 Pro.
Chief Executive Officer, PT Momentum Teknodata Semesta
3wThe shift is not in capability, it is in where reasoning sits in the workflow. When planning moves upstream of generation, the model stops being a tool for answers and starts acting as a system that shapes how problems are approached. That changes build quality more than any benchmark delta.
ရွမ္ျပည္နယ္လာရီူးsoeLinshwehmonea
And one day, in the year 9001, we'll be able to try it without swiping the card.
Google Gemini is an amazing AI assistant
Congratulations for the release. Gemini-3.1-pro shows an improvement in grounded hallucination rate relative to Gemini-3-pro (10.4%) on Vectara hallucination leaderboard https://github.com/vectara/hallucination-leaderboard