Overview
Who this competition fits
Research-level teams interested in fluid intelligence and abstract reasoning tasks, especially people comfortable designing custom solvers and running long Kaggle notebook evaluations.
Read the original official blurb
Create an AI capable of fluid intelligence
Resources
What this competition gives you
Taken from the competition's official page.
Preparation
From registration to a first submission
- 01
Python
- 02
Kaggle notebook submission workflow
- 03
abstract reasoning or program-search experience
- 04
custom evaluation debugging
- 05
ability to run long CPU/GPU experiments
Before you commit: The task is a frontier reasoning benchmark with a custom metric, one-submission-per-day cadence, kernel-only submissions, and long runtime limits. Simple supervised baselines are unlikely to teach much without substantial reasoning-system design.
Source
How this page was assembled
Competition information is structured from the official page. Scores are platform estimates for decision support; official rules take precedence.
- Official competition page
- Kaggle
- Last checked
- Mar 25, 2026