The challenge is the peer review system - not saying it can't be done, but facilitating quality reviews is often costly.
There has, however, been a push to publish articles as "open access" which costs more for the author but makes it publicly available free of charge to read.
Overall the system is still a pretty big scam, but would be difficult to make 100% free.
There's groups doing free peer review in a mutual aid sort of way. As I understand it, reviewers don't really get paid anyway and the work is often dumped on students around the professor. Example of a group doing community peer review: https://archaeo.peercommunityin.org/PCIArchaeology/
The way I see it, this "free" journal is gonna have some overhead, from servers to maintainers, coordinators, and potentially even designers to help get consistency.
Some people may be able to support with their free time, but ultimately if those people/systems are going to be paid, the platform will need a revenue stream, and like magic we're back to square one, albeit with hopefully significantly lower profit margins.
For any one journal, very few, maybe even fractions of a headcount per journal, but for the thousands of journals out there spanning dozens of disciplines and hundreds of specialties, it adds up. If you want to make the end-all-be-all magic journal of all-topicness and maintain a respectable level of quality, you're going to need quite a few SMEs policing the submissions.
There's millions of scientific papers published annually - you need people to process all of that information and moderate peer reviews.
Ok, but we're talking about thousands of dollars in fees for a single journal. There's no reason that a single journal should have costs anywhere near thousands of dollars for a single article.
The average number of articles published per journal per year is ~110. Let's say a major journal publishes probably closer to 300/yr.
Assuming you try and barebones it with 3 staff members, a technical lead for screening, a graphics / visual editor, and a peer review manager. Assume you want someone relatively competent for your journal so you pay each (inclusive of overhead & benefits) ~$150k/yr.
$150k/yr × 3 / 300 articles = $1.5k/article
Again, not saying it's a perfect system and things can definitely benefit from economies of scale, but it really doesn't take much to get $1k/article in expenses to pile up.
I'm not convinced that 3 full time staff is barebones given that the writing and formatting is being done by the authors, and that a solid chunk of what normally falls under the editing umbrella is being done by peer reviewers who are also unpaid.
Even if that is a fair representation of the cost to the journal to get the article published, that would mean they would break even, maybe even earn a profit purely on the submission fees. Never mind that multiple universities pay hundreds of thousands of dollars to subscribe to the journals.
States/academic institutions have to make it part of the job description of people. Get designated an editor of a journal? Your Uni understands and hands you an additional TA to lighten the load elsewhere and/or deal with the paperwork aspects.
The reviewing itself is already done pro bono anyways.