A closely followed measure of artificial intelligence token prices has fallen below $1 per million tokens, according to a CNBC item summarized by Techmeme. The LLM Token Expenditure Index’s average cost per million tokens has dropped to $0.97, CNBC’s Alex Harring reports, down from a reported high of $2.07 on May 28. Techmeme’s summary characterizes the move as part of a sharp months-long decline. The provided item does not specify which models, providers, or workloads are included in the index, nor does it give a stated cause for the decline. That matters: token-price benchmarks can move because of provider price cuts, shifts in model mix, changes in usage patterns, or methodology, but none of those explanations is established by the supplied reporting. For now, the concrete takeaway is narrow but useful: one tracked average for LLM token costs is less than half its late-May high. The story should be treated as a pricing signal rather than a complete read on AI infrastructure economics until more detail is available on the index construction and the drivers of the move. Who benefits: Builders and operators paying for LLM usage by the token may benefit if the index reflects prices they can actually access. Investors get another datapoint suggesting AI software cost curves are still moving quickly. Who's exposed: Model providers and infrastructure companies could face pricing pressure if falling token costs reflect broader market behavior. The supplied item does not identify which providers are most affected.