Unused prepaid credit is refundable for 30 days from purchase. Contact support@unorouter.com for refund requests.

UnoRouter

Open source LLM gateway with failover and one API key

UnoRouter is a self-routing LLM gateway for product teams that sends each request to a healthy model provider and keeps one API key for all of them.

Visit website

Community listing

Listing status

25upvotes

Community score

—

Founder

Compared against 5 alternatives

Market comparison

Launched on Launch Llama Aug 10, 2026 · #58 of 108 that day

About

UnoRouter is a self-routing LLM gateway for product teams that sends each request to a healthy model provider and keeps one API key for all of them. It speaks the OpenAI API, so the same key serves models from OpenAI, Anthropic, Google, Meta, DeepSeek, Qwen and Mistral.

If a provider fails or slows down, automatic failover moves the call elsewhere and your feature stays up. Teams use it from backend code and from agents such as Claude Code, Cline and OpenCode.

Everything is open source, the chat is local-first, free models are included, and billing is a free tier followed by pay-as-you-go with no subscription.

Follow UnoRouter: X (Twitter)

Use Cases

Moving an existing OpenAI-based app to Claude, Gemini or DeepSeek by changing only the model name
Keeping a production chatbot online with automatic failover between providers
Giving a coding agent every major model through one OpenAI-compatible base URL

Who it is for

Developers building apps or agents who want one OpenAI-compatible API across many AI providers.
Users of AI coding agents and terminal CLI tools who want access to many models with one key.
Character roleplay and fiction chat users who import cards into SillyTavern, Janitor.AI, or similar clients.

Pricing

Free tier with free models to start, then pay-as-you-go from a $1 top-up. Credits never expire and no subscription is required.

View UnoRouter pricing →

Verdict

Worth it if you call more than one model provider, or expect to. The single OpenAI-compatible endpoint means switching models is a string change rather than an integration rewrite, and automatic failover keeps a provider outage from taking your app down with it. Being open source and self-hostable, there is no lock-in: you can run the same gateway on your own infrastructure. Less compelling if you are committed to exactly one provider and have no plans to compare alternatives.

FAQ

How does UnoRouter pricing work?

You pay per token consumed. Each model has its own input, output and cache rate, billed from your account credit. There is no monthly subscription required for Pay As You Go.

Do I need a subscription?

No. Pay as you go is fully self-serve: top up from $1, use any model, and the credit never expires.

Is the API compatible with OpenAI?

Yes. UnoRouter exposes an OpenAI-compatible endpoint for chat completions, plus Anthropic and Gemini compatible endpoints. Point any OpenAI SDK at our base URL and use your UnoRouter API key.

Which models are included?

Every plan includes access to all supported models (Claude, GPT, Gemini, DeepSeek and more). We do not gate models by tier.

What happens if a provider fails?

UnoRouter routes requests across multiple providers with automatic failover. If one provider is rate-limited or down, we retry against the next healthy upstream transparently.

Can I get a refund?

Unused prepaid credit is refundable for 30 days from purchase. Contact [email protected] for refund requests.

Can I use it with coding agents like Claude Code, Cline or OpenCode?

Yes. Add UnoRouter as a custom OpenAI-compatible provider with its base URL and your key, then pick any model by name.

How long can my conversations or character history be?

UnoRouter supports the full context window of every model, up to 1M tokens on Gemini and 200k on Claude. Long character conversations and long codebases both work.