Anthropic releases its flagship large language model, Claude Opus 4.8.

Anthropic has released Claude Opus 4.8, its flagship model. The model boasts significant improvements in terminal engineering and knowledge work capabilities, surpassing the unreleased Mythos in some metrics. A core improvement is in honesty: the code defect false negative rate has been reduced to 1/4 of the previous generation, and the probability of overconfidence has been reduced to 1/10. Both Cursor and the Devin CEO have given it high praise.