Open Models Change The Economics of AI
- Source
- youtube.com
- Author
- Y Combinator
- Date
Context
Jeffrey Morgan, co-founder and CEO of Ollama, a for running open AI models locally and in the cloud, told Y Combinator's Light Cone podcast that cost is by far the biggest reason businesses adopt open models, with customization their longer-term aim. Morgan's evidence is Ollama's own data on average weekly use per developer on its cloud service (tokens are the chunks of text a model processes). It jumped first when open models could power coding agents, then in April when OpenClaw let non-developers hand long tasks to open models.
Morgan said windows, the amount of text a model can consider at once, grew from 128K to over a million, and put token use through Ollama's cloud at about 150 times its level at the start of the year. These are company figures from its own users, not the whole market. The AT&T example, 40 percent of token use moved to open models, was relayed from a news article, not Ollama data.
- Tool — An operation that the software around a model makes available for the model to request, with a name, expected inputs and a kind of result.
- Token — A piece of text, such as a word, part of a word or a punctuation mark, that a language model represents as numbers and handles as a unit.
- Context and context window — Context is the information available for a generation; the context window is the capacity available to hold it.
- fine-tuning — Taking a trained model and training it a bit more on your own examples so it gets better at one specific job.
Checking sign-in…
Loading comments…






