How to use More AI for free: OpenRouter’s free models and other zero-cost routes
You do not need a paid API account to start. Three routes cost nothing: OpenRouter’s :free models, the free tiers at Google AI Studio and Groq, and models running on your own machine. Here is how to set each one up and where each one runs out.
The More AI team · Updated August 21, 2026
What “free” actually means here
More AI talks to model providers with a key you supply, so “free” is never a plan we sell you — it is a provider that does not charge for a given model, or a model running on your own computer. Three routes cost nothing today, and the app lists them together at the top of setup under “Start free”.
They are not equivalent. A local model costs nothing and never leaves your machine, but it is only as good as your hardware. A provider’s free tier gives you a strong hosted model with a daily cap. Pick by what you are doing, not by what sounds best.
- Local models via Ollama or LM Studio — no account, no key, no limit other than your own hardware
- Google AI Studio — a free key with a daily allowance, no card required
- Groq — a free tier on open-weight models, and the fastest responses of the three
- OpenRouter — a free key that reaches every model whose id ends in :free
The fastest route: an OpenRouter key and the :free models
OpenRouter is one key that reaches models from every major lab, and a subset of them is published at no charge — their ids end in “:free”. Creating the key takes about two minutes and needs no card.
The free variants are rate-limited rather than metered: at the time of writing OpenRouter allows 20 requests per minute and 50 requests per day on :free models, rising to 1,000 a day once an account has ever purchased $10 of credit. Those numbers change, so check the limits page before planning around them.
- Create a free account at openrouter.ai and open the Keys page.
- Press “Create Key”, give it a name, and copy the value — it starts with sk-or-.
- In More AI open Settings → Providers (or the “Start free” group during first-run setup) and pick OpenRouter.
- Paste the key. The app immediately fetches OpenRouter’s live model list.
- In the model picker, search for “:free” to see only the models that cost nothing.
- OpenRouter — create an API key — The page you copy the key from. A free account is enough.
- OpenRouter — models list — Filter by price to see which models are currently free.
- OpenRouter — rate limits — The official daily and per-minute caps on :free models.

Google AI Studio and Groq
Both hand out a key on a free tier with no card. Google AI Studio is the more generous of the two on daily volume and its Flash models handle long documents well. Groq runs open-weight models on its own hardware and answers noticeably faster than anything else here, which makes it pleasant for back-and-forth work.
Adding either is the same three clicks as OpenRouter: create the key, open Settings → Providers, paste. You can hold keys for all of them at once and switch model per conversation — there is no “plan” to change.
- Google AI Studio — get an API key — Free tier, no card. The key starts with AIza.
- Gemini API — current rate limits — What the free tier allows today, per model.
- Groq Console — API keys — Free tier on open-weight models, very fast responses.
Free and offline: models on your own machine
Ollama and LM Studio run a model on your own computer and expose it over a local OpenAI-compatible server. More AI connects to it with no key at all — the “Connect” button probes the local server and that is the whole setup.
This is the only route with no daily cap and no request leaving your machine, which is what makes it the right one for confidential documents. The trade-off is hardware: a 7–8B model runs comfortably on a modern laptop and is fine for drafting, summarising and simple edits; anything approaching frontier quality needs a lot more memory.
- Ollama — download — Install, run “ollama pull qwen3”, and More AI finds it.
- LM Studio — local server docs — The same thing with a graphical model browser.
Where free runs out — and what to do then
Free tiers are capped by day, not by task, so they run out exactly when work gets interesting: a long agent run with many tool calls can spend a day’s allowance in one sitting. Local models have no cap but hit a quality ceiling on hard reasoning and long code changes.
The honest answer is that free is for learning the tool and for light daily work. When you outgrow it, you are not forced onto our plan — you top up the same provider key you already have, and the app keeps working exactly as before. That is the point of bringing your own key: the upgrade is a decision about a model, not about us.
Frequently asked questions
Do I need an account with More AI to use free models?
No. The desktop app runs on your own provider key and works without signing in. An account is optional and comes in when you want the cloud: syncing between devices, opening your work in a browser, and runs that continue while your computer is off.
Which free option should I start with?
If you want the widest choice of models, OpenRouter — one key reaches every model whose id ends in :free. If you work with long documents, Google AI Studio. If nothing may leave your machine, Ollama.
How many free requests do I get on OpenRouter?
At the time of writing, 20 per minute and 50 per day on :free models, rising to 1,000 a day for accounts that have ever purchased $10 of credit. Check OpenRouter’s limits page for the current figures — they change.
Is a free model good enough for real work?
For drafting, summarising, questions and light editing, yes. For long multi-file code changes and hard reasoning, a paid frontier model is still noticeably better. You can switch per conversation, so use the cheap one by default and the expensive one when it matters.
Do free models see my files?
A hosted model sees whatever you send it, the same as any API. Read each provider’s terms on whether free-tier traffic may be used for training — that policy differs by provider and by tier. A local model through Ollama or LM Studio sends nothing anywhere.