← All articles

Coding

Google DeepMind's Gemini 3 Pro Hackathon: Details & Overview

3 min read

Updated

Discover details about Google DeepMind's Gemini 3 Pro Kaggle Hackathon, featuring a $500,000 API credit prize pool and advanced multimodal features.

Google DeepMind launched a developer hackathon challenging participants to build multimodal applications using the Gemini 3 Pro model within Google AI Studio.

  • Prize Pool: $500,000 in Gemini API credits.
  • Focus: Reasoning, multimodal inputs (video, code, images), and agentic prototypes.
  • Platform: Hosted on Kaggle.

The event encourages developers worldwide to build working prototypes that combine reasoning, code execution, and multimodal inputs, moving beyond simple data-science notebooks or basic text experiments.


Official Hackathon Details

The official schedule and guidelines for the competition include:

FieldDetail
Start DateDecember 5, 2025
Submission DeadlineDecember 12, 2025
Prize Pool$500,000 in Gemini API usage credits
Submission RequirementsProjects must use Gemini 3 Pro in Google AI Studio. External tools, datasets, or APIs are allowed as long as the core logic runs through Gemini 3 Pro.
Allowed Domains & ThemesScience, health, education, accessibility, business, technology, and other creative fields.

Gemini 3 Pro Core Features

Gemini 3 Pro is the underlying model powering the hackathon. It introduces several key capabilities:

  • Multimodal Input & Output: Native support for processing and generating text, images, video, code, and mixed-media inputs.
  • Advanced Reasoning & Coding: Combines logical reasoning with code generation and media handling, enabling complete application workflows rather than basic conversational outputs.
  • Enterprise Integration: The model is also available for enterprise platforms and developer environments seeking built-in coding and reasoning capabilities.

Early Activity & Project Prototypes

Early developer activity on Kaggle and community forums indicates strong interest in testing the boundaries of the preview model.

Shared Notebooks and Builds

The Kaggle competition code repository contains public notebooks demonstrating early experimentation with the preview models. These notebooks show participants building workflows that combine data processing with media and interface handling.

Community Projects

Community reports highlight several prototype concepts built using Gemini 3's multimodal features:

  • Interactive 3D Screens: Particle displays controlled via hand gestures.
  • Sign-Language Tutors: Real-time video translation utilities with confidence scoring.
  • AI-Powered Media Tools: Automated podcast hosts and context-aware image editing and panorama tools.

How to Get Started

For developers interested in exploring Gemini 3 Pro or entering similar competitions:

  1. Register: Visit the official competition platform on Kaggle to sign up and view the rules.
  2. Access the Model: Open Google AI Studio and select the Gemini 3 Pro preview model.
  3. Define the Scope: Focus on a project that utilizes multimodal inputs (like video or images) combined with reasoning or code execution.
  4. Build a Prototype: Prioritize core functionality and demo potential over visual polish.
  5. Submit: Upload the required deliverables, including the execution notebook or code repository and a demo link or video.

Even for teams that do not win the top prizes, the competition offers a structured way to experiment with next-generation model APIs at no initial infrastructure cost.


FAQ

What is the Gemini 3 Pro hackathon?
It is a developer competition organized by Google DeepMind and hosted on Kaggle, designed to showcase the multimodal and reasoning capabilities of Gemini 3 Pro in Google AI Studio.

What are Gemini 3 Pro's main features?
Gemini 3 Pro features native support for mixed-media inputs (text, image, video, and code), improved logical reasoning, and enhanced code generation.

What types of applications can be built?
Projects range from document parsing and data tools to interactive media applications, gesture-controlled interfaces, and education software.


Public Resources & Reference Links