BLOG POST

For most of the last six years, the most capable LLMs were locked behind APIs. You could call them, you could build on them, but you couldn’t look inside, customize the weights, or run them on your own infrastructure. That was simply the reality of working with frontier models. In