What is this website?
UnoRouter routes hundreds of AI models via one OpenAI-compatible API key for apps, agents, and on-site chat. Free models available; paid models billed by usage.
Key features
- Offers OpenAI-compatible endpoints, so switching models is usually a model-name change
- Lists free and paid models together with context, unit prices, and capability labels
- Can fail over across provider groups when an upstream is rate-limited or down
- Lets each API key set quotas, model allowlists, IP limits, and provider-group pinning
- Includes a browser character chat with card import; conversations stay local by default
- Documents setup for coding agents, character clients, chat apps, and CLI tools
Common use cases
- Developers point an OpenAI SDK base URL to UnoRouter and reuse the same code across upstream models
- Users of Cline, Roo Code, Claude Code, and similar coding agents switch between free and paid models with one key
- Character clients such as SillyTavern or Janitor.AI paste the compatible proxy endpoint and pick a model
- People who do not want to configure a key first open the on-site chat and talk through wired free models
- Small teams review request logs, balances, and charges in the console when auditing per-key usage
Who is it for?
- App and agent developers who want multi-model access with minimal code changes
- Individuals who need flexible model switching inside coding agents or character clients
- Testers who validate a flow on free models before moving to paid usage
- Operators who assign quotas and access limits across different API keys
- Poor fit: teams that only need one official provider SDK, or that must self-host all inference with no third-party routing
How does it compare with similar tools?
Use these checks to decide fit:
- Interface shape: whether you need OpenAI compatibility and one setup for many models
- Cost structure: whether you need usable free models plus pay-as-you-go paid models
- Client coverage: whether coding agents, character chat, and generic chat clients all matter
- Control detail: whether key limits, model allowlists, or provider-group pinning are required
- Chat location: whether you need key-free on-site chat or only a pure API
If you stay with one official SDK and do not plan to switch models, the provider API is usually simpler. If you need multi-upstream switching while reusing existing clients, a gateway-style product is a closer match.
FAQ
Q: Is UnoRouter free?
A: Models with a :free suffix do not debit balance, but they have rate and upstream capacity limits. Paid models bill per token or per call; credit comes from top-ups or subscriptions.
Q: How do I connect an existing app?
A: Set the OpenAI-compatible client's base URL to https://api.unorouter.com/v1, add an API key from the console, then choose the target model name.
Q: How does on-site chat differ from the API?
A: On-site chat can start on free models without signup, and chats stay in the browser by default. API requests always need a key and suit apps, agents, and third-party clients.
