All this will be an interesting side note in the history of language models in the next eight months when roughly 1.5 billion iPhone users will get a local language model tied seamlessly to a mid-tier cloud based language model native in their OS.

What I think will be interesting is seeing which of the open models stick around and for how long when we have super easy ‘good enough’ models that provide quality integration. My bet is not many, sadly. I’m sure Llama will continue to be developed, and perhaps Mistral will get additional European government support, and we’ll have at least one offering from China like Qwen, and Bytedance and Tencent will continue to compete a-la Google and co. But, I don’t know if there’s a market for ten separately trained open foundation models long term.

I’d like to make sure there’s some diversity in research and implementation of these in the open access space. It’s a critical tool for humans, and it seems possible to me that leaders will be able to keep extending the gap for a while; when you’re using that gap not just to build faster AI, but do other things, the future feels pretty high volatility right now. Which is interesting! But, I’d prefer we come out of it with people all over the world having access to these.