Freedom through free tools.

A confession, first

I have a big PC in my basement. It can run serious AI models locally, and I pay for the expensive subscriptions on top of that. I am, fully, one of the people the current setup works for.

That’s exactly why it bugs me. None of what I build with that gear actually requires it. Almost nobody says that out loud. If you want to build things with AI and you don’t have a basement PC or $200 a month to spare, the world quietly lets you believe you’re locked out. You’re not.

The AI doesn’t run on your computer

It runs on someone else’s big machine. Yours sends the question and catches the answer. And the companies running those big machines each give away a real free slice, every day. Free doors into the same brains people pay hundreds a month for.

Which means the machine you already own is enough. A $200 used MacBook Air. A $20 Raspberry Pi, the “almost a computer.” The phone in your pocket. Your machine is the walkie-talkie, not the brain, and walkie-talkies are cheap.

So why doesn’t everyone know this?

Because the information is gatekept. Not by a villain. By an attitude. The facts about free access sort of exist. Scattered across forum threads and dev blogs, wrapped in jargon, half of it out of date, delivered with an if-you-have-to-ask shrug. The people who would benefit most never even hear that the doors exist. Students. Tinkerers. Writers. People on old laptops and tight budgets.

Gatekeeping is hoarding the ability to figure it out. ModelHubby ships that ability whole.

That’s the whole ethos, in two lines: tech accessibility for all, and platform agnosticism. No lock-in. No favorites. Use the free doors. Use paid providers. Run models on your own machine. Mix all three, in whatever order you choose, with the free options as your safety net when anything else fails. We keep the map current either way.

What free always includes

We hold ourselves to one test, written down where we can’t un-write it:

The determined do-it-yourselfer can always succeed end-to-end, free, unaided. Paid removes time and adds certainty.

Concretely, free, forever, includes:

• The entire open directory: every free place we track, the real limits in the provider’s own words, the sign-up friction, whether they keep and train on what you type, and honest tombstones when a provider dies or goes paid.
• The open-source tools: the cascade builder that lines up your free options best-first and hands off gracefully when one runs out, and the local machinery that makes it work.
• Enough full guides to do it yourself, including privacy-clean picks. Privacy is never paywalled here.
• Weekly updates and a public changelog. No blur, no teasers pretending to be content.

If you’re determined and have the time, you never need to pay us a cent. That is a design decision, not an oversight.

What paid is honestly for

Paid (Indie, $12 a year) buys exactly two things: time and certainty. Daily verified updates instead of weekly. Alerts the moment a provider changes its terms. The full VaiB & HaiD library: every guide, every runbook, the “which brain for which job” editorial. The done-for-you exports, generated for your exact setup instead of hand-assembled from the free docs.

And one more thing, the part I care about most: paid memberships fund free tools for everyone. That is not a promise about the future. vaibos.com/report is live, free, and funded exactly this way. The cascade idea itself was born building revelocity.co. Every paid member keeps the free floor free, and growing.

The short version

The good AI has free doors. You can walk through them on any old machine. The only thing standing between the people who could use that and the people who know it is gatekeeping. So VaiB scouts the doors, HaiD makes the setup last, and we publish the map.

Go see the free options, or start with a guide. And if this is the thing you needed and didn’t know existed, tell the next person. That’s how gates come down.

— Peter, ModelHubby operations