ARC Prize
ARC Prize builds scientifically grounded AGI benchmarks that measure the gap between human and machine intelligence. Its ARC-AGI benchmark series is trusted by leading AI labs and academic researchers, with ARC-AGI-3 standing as the world's only unbeaten benchmark measuring agentic intelligence. The organization also runs public competitions via Kaggle, offering over $2M in prizes for open-source progress.
- Artificial Intelligence
- Data & Analytics
- Software Only
Funding
Raised to date
$500KRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Founders
Founder details are not available yet.
Product
Problem
Conventional AI benchmarks often fail to distinguish between systems that memorize patterns and those that genuinely learn, making it difficult to track real progress toward artificial general intelligence (AGI). As AI capabilities advance, existing evaluation methods provide noisy signals and do not reliably measure fluid intelligence or the gap between what is easy for humans and hard for machines.
Solution
ARC Prize builds low-noise, high-signal AGI benchmarks that scientifically measure the gap between human and machine intelligence. The ARC-AGI benchmark series is trusted by leading AI labs and academic researchers and has repeatedly identified key inflection points in AI progress, including the emergence of reasoning systems and capable AI agents. ARC-AGI-3 currently stands as the world's only unbeaten benchmark that measures agentic intelligence. Beyond benchmarking, ARC Prize runs public competitions—offering over $2M in prizes via Kaggle—to reward open-source novel progress and accelerate research toward matching human learning efficiency.
Target Audience
Primary customers are AI research labs, academic institutions, and standards organizations such as NIST that need rigorous, scientifically grounded benchmarks to measure progress toward AGI and evaluate AI system capabilities.
Features
- ARC-AGI benchmark series trusted by leading AI labs and top academic researchers for measuring fluid intelligence
- ARC-AGI-3, the only unbeaten benchmark designed to measure agentic intelligence
- Public leaderboards comparing model performance against human baselines on benchmark tasks
- Interactive task interface letting users play benchmark tasks and track human completion data
- Prize competitions in partnership with Kaggle, awarding over $2M for open-source novel progress
- Benchmark data provided to NIST's Center for AI Standards and Innovation (CAISI)