I’m sure OpenAI or Meta or Apple could have created something like this (or better) even sooner if it was their goal, but their business model makes them unwilling, regardless of their unlimited resources and world-class talent.

A few months ago, I thought it would be impossible to ever run an LLM of this quality on my hardware. Given how much value they create for tech companies, I’m not surprised that an open-source project is the first to put these tools directly in the hands of users on a budget.