Sign into an existing account

We have a true open source LLM with o1 performance (or better?). Trained on 671B params with reinforcement learning. Cost per million tokens is ~$2, compared to ~$60 from OpenAI. Mixture-of-experts MOE architecture (similar to model orchestration?).

I still need to read the white paper but what should I be looking for with this? LLM deployment/dev becomes super cheap, all the major US companies need to cut prices to compete, people can run a much improved LLM on their home PC, I really just see this as cutting costs, increasing competition and a good thing for people, despite the losses for corporate America tech(currently in a tail-spin). Security is a massive concern being it is a Chinese product so we will need to make a more secure version that can be trusted. Also, the compute power would be largely distributed meaning infrastructure strain would be significantly less than an enterprise focused ecosystem.

Trying to get my head around this, any ideas? Boomers, no one gives a shit about how you are retired and hate new technology, we know.