What if AI worked at 1.000.000 tokens per seconds?

The article explores a thought experiment about what artificial intelligence capabilities would look like if a model could process one million tokens per second. It clarifies that this metric could mean different things—context capacity, input processing speed, aggregate system throughput, or single-agent output speed—and uses interactive examples to show how various AI workloads (generating story endings, drafting apps, critiquing candidates) would complete at dramatically faster speeds, from seconds to minutes instead of hours.