UnoRouter: the open source OpenRouter alternative
OpenRouter is closed source. UnoRouter does the same one key, 200+ model job with the entire stack public under OSI licenses, self hostable, and with a hosted free tier to test on.
UnoRouter is an open source OpenRouter alternative: the same one key for 200+ models, with the whole stack public under OSI licenses at github.com/unorouter. You can read how prompts are handled, self host the gateway, or use the hosted version, which serves most models on a free tier of about 1 request per minute per model and bills paid use pay as you go.
Searching for an open source OpenRouter alternative usually means one of three things: you want to read the code that handles your prompts, you want to run the gateway yourself, or you want out of a black box. UnoRouter covers all three. Same one key, 200+ model setup, whole stack public.
Why open source matters in a gateway
A gateway sits between your app and every model you call, so it sees every prompt and every reply. With a closed gateway you trust a privacy page. With an open one you can check what gets logged, how keys are stored, and how requests are routed before anything sensitive leaves your machine.
What is actually open
Everything is public on GitHub under OSI licenses: the web frontend with the chat client, the gateway itself, the Discord bot, and the provider sync that keeps the catalog fresh. This is not a thin SDK wrapped around a closed core. The running product is the repo.
Against the usual picks
OpenRouter has the widest catalog, but every line of it is closed. LiteLLM is genuinely open and a solid self hosted proxy, but it is bring your own keys plumbing: you still manage provider accounts and there is no hosted catalog. UnoRouter sits between the two, open like LiteLLM, hosted with one key and 200+ models like OpenRouter.
Self host it or just use it
Clone the repos and run the stack yourself, or point any OpenAI compatible client at https://api.unorouter.com/v1 and use the hosted version. The hosted free tier covers most of the catalog, so you can test either path before committing to it.
The short version
If all you need is local routing over your own provider keys, LiteLLM is the leaner tool. If you want an OpenRouter style hosted gateway where you can read, audit, and fork the entire stack, UnoRouter is the one that ships its source.
Ready to try it? Create a free account or browse the models.
Frequently asked questions
Is UnoRouter really fully open source?
Yes. The web frontend with the chat client, the gateway, the Discord bot, and the provider sync are all public under OSI licenses at github.com/unorouter. The hosted product runs from the same code.
Can I self host UnoRouter?
Yes. Clone the repositories and run the stack yourself, or point any OpenAI compatible client at the hosted endpoint https://api.unorouter.com/v1 and skip the ops work.
How is UnoRouter different from LiteLLM?
LiteLLM is a self hosted proxy where you bring provider keys and manage accounts yourself. UnoRouter is hosted with one key, 200+ models, and a free tier, while staying open source like LiteLLM.
Wearing the UnoRouter Discord tag shortens your free-model rate limit window by 25 percent, so a model you could call once a minute you can call every 45 seconds.
Nevika takes any OpenAI-compatible endpoint as a custom proxy. Here is the one minute setup, which model to pick, and the two errors people hit.
A roleplayer ranks three Claude Opus versions on UnoRouter: memory, emotional depth, drive, and creative writing. 4.8 wins, 4.6 beats 4.7, and each one wins or slips in a clear place.