“Closed models dominate, with on average 80% of monthly LLM tokens using closed models despite much higher prices - on average 6x the price of open models - and only modest performance advantages. Frontier open models typically reach performance parity with frontier closed models within months, suggesting relatively fast convergence. Nevertheless, users continue to select closed models even when open alternatives are cheaper and offer superior performance. This systematic underutilization is economically significant: reallocating demand from observably dominated closed models to superior open models would reduce average prices by over 70% and, when extrapolated to the total market, generate an estimated $24.8 billion in additional consumer savings across 2025.”
This is another sign that the AI bubble almost certainly has to pop. But there’s an interesting implication here. Will open-source AI inherit the future?
Linux, Android, MySQL, Git, WordPress - are just a few of the open-source software solutions that dominate modern software & the internet. Will the bedrock of 2030s AI be open-source?
I’ve found that a lot of the people who complain the loudest about the costs of AI also seem to refuse to believe that local AIs are even possible. Quite frustrating.
I honestly think a big part of it is the tendency to believe that the big tech companies aren’t just extremely wasteful - basically thinking that if it could be provided cheaply for 99% of the benefit, the tech companies would be doing it themselves.
I really like the idea of people running their own LLMs. The cost is at the point of use, and it’s only giving responses that are explicitly asked for.