Meta’s new plan for AI: what you actually need to know

Meta has released a new, smaller AI model called Muse Glimmer and promised to release a more powerful one called Muse Spark 1.2 in the coming weeks. Unlike the major competitors, which keep their AI locked behind their own websites, Meta is releasing these as open-weight models. This means anyone with the right computer hardware can download the AI, run it locally on their own machine, and modify it however they like without needing to pay Meta for every use.

To understand this, it helps to think of an AI model as a highly trained apprentice. A typical company trains its apprentice in a private, high-security room and only lets you interact with the apprentice through a window—this is the closed, proprietary model. You send a question, they send an answer, and you pay a fee for the service. Meta is taking a different approach: they are effectively printing the apprentice’s entire knowledge base onto a portable manual and giving it to you for free. You can take this manual home, load it onto your own computer, and use it as much as you want without asking the original company for permission. Because the code is open, other developers can also take Meta’s model and build their own custom versions or specialized tools on top of it.

This change tells us a lot about where the AI industry is heading. Meta is trying to carve out a spot as the accessible, flexible alternative to the massive, expensive systems developed by its competitors. By making these models easier to run locally, Meta hopes to stop relying on the cloud services of big labs and instead put AI directly on your laptop or phone. However, the company’s vision—which imagines these tools managing our relationships, teaching us new skills, and automating our daily hobbies—raises a deeper question. We are moving toward a world where technology can do more for us, but in the process, we have to decide what parts of our lives are actually worth handing over to an assistant. If we use AI to write our emails, learn our languages, and pick our recipes, we might get more done, but we also lose the personal effort that often makes those activities meaningful in the first place.