Research

AGI Research

Inpiring Open Source Progress

Welcome, AI researchers!

If you're an undergrad, graduate, PhD student, or simply curious about the frontiers of artificial general intelligence (AGI), the ARC Prize and Foundation invites you to join one of the most impactful and exciting research communities today.

Technical Guide

Learn the ARC-AGI task format, evaluation criteria, and how to build and submit your solutions.

Research Grants

Need funding to support your research? We may be able to help.

Partner Resources

Our partners offer compute credits and tools to help enable your ARC-AGI research.

Why Research ARC-AGI?

Artificial General Intelligence represents potentially the most transformative breakthrough in human history - AI capable of inventing, reasoning, and discovering alongside us. At the ARC Prize Foundation, our benchmarks highlight the critical gap between what is "feasible for humans" and "hard for AI," guiding researchers toward genuinely generalizable solutions rather than superficial performance improvements.

Your research into ARC-AGI directly influences the future of AI development. By contributing, you'll help guide global AI research and gain recognition from top academic institutions and industry leaders.

Get Started with ARC-AGI

Begin your journey by understanding the foundations laid out by François Chollet’s influential paper, ”On the Measure of Intelligence”. Next, familiarize yourself with the ARC datasets - starting with the ARC-AGI-1 public training set, featuring 400 tasks ideal for initial experimentation. When you feel ready to progress, explore the more advanced ARC-AGI-2 benchmark, designed explicitly to emphasize deeper abstraction and generalization.

Learn from Researchers Who’ve Made an Impact

The ARC Prize 2024 competitions inspired novel research, resulting in over 40 papers submitted. Featured in 2024 was Test Time Tuning, an inference time compute variant.

ARC-AGI researchers frequently gain public visibility, with their work featured in renowned media outlets such as Nature, Time, New Scientist, and Forbes. Previous researchers were covered by Machine Learning Street Talk.

Daniel Franzen and Jan Disselhoff on Machine Learning Street Talk

ARC Prize 2024 Top Score Winners & Paper Prize Runners Up

Daniel Franzen & Jan Disselhoff, the ARChitects

Clément Bonnet on Machine Learning Street Talk

ARC Prize 2024 Paper Winner

Clément Bonnet

Leverage Our Resources and Partnerships

Our goal is to support the research community in the pursuit of AGI. We partner with industry to offer compute credits, specialized datasets and grant opportunities.

Visit our partner page to see how you can leverage these resources for your research.

If you don't see someone you'd like to partner with, please reach out to us (team@arcprize.org) and we'll do our best to make it happen.

Get In Touch

We love hearing from the research community. Our goal is to support the research community in the pursuit of general intelligence. Please reach out and let us know what you are working on. We are always looking for novel ideas to promote within our network.

Plug into the community:

  1. Subscribe to the newsletter
  2. Join the Discord server
  3. Follow on Twitter
  4. Subscribe on YouTube

You can also reach our team at team@arcprize.org or message us on Discord.

Join us in pulling the future forward.