Nearly 10k and 400 stars on those respective repos.
A way to run a large language model on any operating system, in any OS, in a simple, local, and privacy respecting manner?
For linux we have docker, but Windows users were starving for a good way to do this, and even on linux, removing the step of configuring docker (or other container runtimes) to work with nvidia, is nice.
And it's still FOSS stuff they aren't being paid for, currently. But there are plenty of ways to monetize this.
Here's an easy one: tie in the the vpn service they have to allow you to access the web ui of the computer running the llamafile remotely. Configure something like end to end encryption or or nat traversal (so not even mozilla can sniff the traffic), and you end up with a private LLM you can access remotely.
With this, maybe they can afford some actual development on firefox, without having to rely on google money.