This is the first AI related “breakthrough” I can remember being attributed to Apple

Apple researchers have developed a breakthrough “multi-token prediction” framework that enables large language models to generate text up to 5 times faster while maintaining output quality

Has it finally entered the chat?

via Perplexity