Author(s): Eivind Kjosbakken

Originally published on Towards AI.

Discovering Top 3 Frontier LLMs Through Benchmarking — Arc AGI 3

In the last few weeks, we have seen the release of powerful LLMs such as Qwen 3 MoE, Kimi K2, and Grok 4. We will continue seeing such rapid improvements in the foreseeable future, and to compare the LLMs against each other, we require benchmarks. In this article, I discuss the newly released ARC AGI 3 benchmark and why frontier LLMs struggle to complete any tasks on the benchmark.

The article discusses the recent developments in LLM technology and the release of the ARC AGI 3 benchmark, emphasizing the challenges frontier LLMs face in achieving human-level performance on benchmark tasks, with many models achieving scores as low as 0%. The author explores several factors contributing to these low scores, including the absence of information during tests, the mismatch between training data and the benchmark tasks, and the concept of benchmark chasing—where model performance is optimized for benchmarks rather than genuine intelligence. The conclusion highlights the hope for future improvements in LLM performance on ARC AGI 3, paired with an emphasis on understanding intelligence without the constraints of benchmarks.

Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI