Claude Opus 5 crushes AI intelligence benchmarks

Anthropic just released Claude Opus 5 and it outperformed every other AI model on the market. It scored 30.2 percent on a tough test called ARC-AGI-3. This score is nearly four times higher than the previous record holder, GPT-5.6 Sol.

The developers of the test were stunned by the results. They noticed the AI created its own math equations to solve problems. They had never seen another model figure out these steps on its own before.

Experts believe this proves the model has better logical reasoning skills than older versions. It is a major jump forward for AI technology. We are watching to see how this changes what these tools can actually do.