Silicon Data's LLM Token Expenditure Index stagnates at 1.99, signaling a slowdown in enterprise migration to premium closed models
Weisenthal argues cheap closed models handle standard developer workloads.
Weisenthal argues cheap closed models handle standard developer workloads.
Commentary on X
Highest rankedI think some people (obviously not the folks at @Silicon_Data) have some mental model of open models as being freely usable open source software like Linux. But they still need costly chips and electricity to use. (Though SOTA open models are clearly cheaper on Chinese clouds)



