AI inference is different. You get the outcome by passing text through some weights at the time you need it. There's no ongoing work besides training and releasing new models. If I had something that rivalled Opus 4+ I could use locally, I would switch in a heartbeat.
No comments yet.