There simply isn’t any value in the models and only in the compute to run them and potentially some higher-order coordination layers.
That’s very bad for the big model labs that banked their future on the opposite, but there just doesn’t seem to be a viable path forward on that approach anymore.
Meta hasn’t been the best global citizen for most of its life, but there may be some mild redemption if it becomes a source of strong open models.
I don't understand how we will sustain an ecosystem of open models like the one that exists today. My thoughts go down this path every time I try to put the pieces together:
- Many open models depend on high quality frontier models as a distillation input
- US closed frontier models drive China's open model strategy despite the business model for open models being somewhat fuzzy
- What is the business model that benefits from open weights, exactly? How do their vendors recoup infrastructure and training costs?
I would be grateful to be enlightened on this topic.
You make money from value-add on top of the models not the models themselves.