Supporting the Future of AGI: RunPod Partners with ARC Prize 2025

The race toward artificial general intelligence isn't just happening behind closed doors at trillion-dollar tech companies. It's also unfolding in the open—in research labs, Discord servers, GitHub repos, and competitions like the ARC Prize.

This year, the ARC Prize Foundation is back with ARC-AGI-2, a new benchmark designed to test reasoning, problem-solving, and generalization in ways that challenge even the most capable AI systems. And with over $1 million in prize money on the line, the competition is open to anyone willing to take a shot at building truly intelligent systems.

At RunPod, we're proud to support this effort by offering $100 in compute credits to all ARC Prize participants.

What is ARC-AGI-2?

ARC-AGI-2 builds on the original ARC benchmark (Abstraction and Reasoning Corpus), a dataset created by François Chollet to test human-like generalization. These tasks are simple for people—like visual pattern recognition, spatial reasoning, and analogical thinking—but notoriously hard for machines.

ARC-AGI-2 ups the ante with a harder, broader benchmark focused on evaluating real progress toward AGI. Solving this benchmark requires models to reason abstractly, generalize from few examples, and apply knowledge in new contexts.

Why It Matters

Unlike traditional benchmarks focused on scale or brute-force pattern matching, the ARC Prize is about human-level reasoning. It's a shot across the bow for teams building smarter, more general AI systems—and doing it in the open.

At a time when much of AI progress is locked behind corporate APIs, the ARC Prize is a rare opportunity to drive progress in a way that’s transparent, reproducible, and accessible to independent researchers.

How RunPod is Supporting the ARC Community

We believe that access to compute shouldn’t be a blocker for good ideas. So we’re offering $100 in free GPU credits to all officially registered ARC Prize participants.

With RunPod, you can:

• Spin up H100s, H200s, A100s, and more in minutes
• Run ARC-AGI-2 benchmarks in a fully configured PyTorch or TensorFlow environment
• Fine-tune models, run training jobs, or benchmark inference across high-performance GPUs
• Save time with persistent storage, prebuilt containers, and easy-to-use CLI tools

Whether you're testing a clever new architecture, running hundreds of simulations, or just need a clean box to run your code—RunPod gives you the power to scale your ideas.

Join the Competition

The ARC Prize is open now, with new tasks rolling out and a vibrant community already experimenting in the open. If you’re interested in participating, visit the ARC Prize website to learn more.

To claim your RunPod credits, apply here.

We’re proud to be a part of this movement—and we can’t wait to see what the community builds next.

Let’s push the boundaries of intelligence, together.