24.9k stars · MIT (read from the About sidebar on 2026-09-10) · v0.9.8 (07 Sep, therefore 7 September 2026 — a past bare date), read from /releases/latest
One local address in front of the free tiers of many AI providers.
▶Repo detailsthe review · specs · pros & cons · install
What it is
FreeLLMAPI is a self-hosted router. Self-hosted means it runs on a computer you control rather than on somebody else's service. It presents one OpenAI-shaped /v1 endpoint and forwards each request to whichever provider you have configured.What it is good for. Anyone doing a lot of cheap, low-stakes model work: testing prompts, bulk classification, drafts nobody will publish. Instead of six accounts and six keys in six scripts, there is one address. It is tiny, about 40 MB of memory when idle, and it runs on a Raspberry Pi. For your own work there is one honest use and one honest warning. The use is experiments — trying a prompt fifty ways costs nothing. The warning is that Grasppy is a product people pay for, and a stack of free tiers has no guarantee behind it, so this does not belong under a paying customer.
- You bring your own accounts and the keys are stored encrypted. The project is not handing you somebody else's access.
- Very small. Node 20 or newer, roughly 40 MB of memory at rest, and it runs on Windows, macOS, Linux or a small ARM board.
- One
/v1address means your existing OpenAI-shaped code needs no changes.
- The documentation says plainly that this is for personal experimentation and asks you not to build anything real on it. Take that seriously.
- No top-tier models, changeable speed, and no service guarantee. Quality drops late in the day as the free daily limits run out.
- The free install gets a monthly snapshot of which models exist. Same-day updates cost $19 a year, or $49 once.
curl -fsSL https://freellmapi.co/install.sh | bash