The good AI has free doors. We keep the map.
Want to build with AI but can’t afford a monster PC or $200 a month? You don’t have to. The companies running the big AI brains each give away a real free slice. ModelHubby tracks every one of them. The limits. The catch. The sign-up steps. Checked constantly, lined up so when one runs out for the day, you glide to the next. On any machine you already own.
FACTS OVER FLUFFAny old machine will do. Really.
The AI doesn’t run on your computer. It runs on someone else’s big machine. Yours just asks it questions. That entry fee everyone assumes? It doesn’t exist.
A $200 used MacBook Air
Or any hand-me-down laptop that still turns on. If it can open a browser, it can talk to the best AI brains in the world.
A $20 Raspberry Pi
The “almost a computer.” It can’t run big AI itself. It doesn’t need to. It just sends the question and catches the answer.
The phone in your pocket
Yes, even that. The heavy lifting happens elsewhere. Your machine is the walkie-talkie, not the brain.
Free runs out. So we planned for that. graceful hand-offs, your order
Every free door has a daily allowance. Use one alone and you hit the wall by lunch. A cascade is the fix. Line up several free options, best one first. When today’s slice of #1 runs out, your work slides to #2 without breaking stride. Then #3. Tomorrow the allowances reset and you start back at the top.
You pick the places and the order. We tell you what each is best at and keep the list current. Already paying for AI? Running models on your own machine? Keep doing that. The cascade makes a fine safety net for the day those fail.
Meet your guides
ModelHubby = “Model Hub” + this married robot couple. They explain which brains are best for writing, analyzing, coding, video. In words that don’t require a computer science degree.
VaiB finds the doors
He scouts every free option and tests that it works today, not last spring. Then he writes the walkthroughs: how to sign up, what it costs you (usually minutes, not money), and what each brain is genuinely good at.
HaiD makes it last
She turns what VaiB finds into something dependable. The cascade order. The graceful hand-offs. The plan for when a provider changes its terms or disappears. He asks “is it true?” She asks “will it keep working?”
No gatekeeping. Ever.
This information sort of exists. Scattered across forums, wrapped in jargon, guarded by a figure-it-out-yourself attitude. We think that’s hoarding. Everything a determined do-it-yourselfer needs to succeed here is free, forever. Paying buys time and certainty, and funds free tools for everyone.
For the technical crowd: the proof
You already know what inference is. What you want to know is why this won’t go stale like every “free LLM API” list before it. Because the registry treats provider facts like infrastructure. Your fallback chain depends on them.
Verified, not scraped
Three checks per provider: record sanity, live model listings, and a tiny real completion. When Google’s model list advertised a model that 404’d on chat, the probe caught it the same day.
“Free” ≠ frictionless
Card, phone, membership, time-to-first-key, and prompt-training policies are first-class fields. Filters never treat “unknown” as a pass.
Quota units, uncoerced
RPM, RPD, tokens, credits, neurons/day, concurrency. Kept distinct, with the provider’s raw wording preserved next to parsed values.
Routing-ready
A local OpenAI-compatible gateway routes model: “auto” across your keys with health-weighted scoring, per-key cooldowns, and circuit breakers.
Provenance on every fact
Records cite their sources with trust tiers and check timestamps. When sources disagree, the conflict is preserved and flagged. Never silently resolved.
Versioned drift tracking
Monthly snapshots and material-change reports: which providers tightened limits, which models vanished, which policies flipped. Two providers are already tombstoned.
Featured free tiers
Ranked views, dossiers, and models for all 48 in the directory.
Groq
Google AI Studio (Gemini API)
OpenRouter (free models)
Cloudflare Workers AI
Cerebras Inference
OVHcloud AI Endpoints
