One API key for OpenAI, Anthropic, DeepSeek, and Google. Transparent pricing, no billing surprises, no outages — plus a free self-hosted model to get started.
from openai import OpenAI client = OpenAI( base_url="https://api.routers.io/v1", api_key="rtr_your_key_here", ) response = client.chat.completions.create( model="gpt-4o", messages=[{"role": "user", "content": "Hello!"}], ) # Same code. Any model. One key.
A Go binary with no runtime dependencies and no database in the hot path. Reliability is the product.
Access OpenAI, Anthropic, DeepSeek, Google, and our free self-hosted models. One key, one format, zero hassle.
Upstream cost plus a small margin. No platform fees, no hidden charges. You see exactly what you pay for.
Go binary, no runtime deps, no database in the hot path. Circuit breakers and automatic failover. Sub-1% error rate target.
Change your base_url and you're done. Works with every OpenAI SDK, LangChain, and LlamaIndex out of the box.
Qwen running on RTX PRO 6000 GPUs. Free tier included. No credit card required to start.
Request logs, cost analytics, and budget alerts. Know exactly what every call costs in real time.
| The problem elsewhere | routers.io |
|---|---|
| OpenRouter: 46+ outages/year | ✓ Circuit breakers, <1% error rate |
| Fireworks rejects 15–23% of requests | ✓ Automatic failover across providers |
| Together AI: surprise bills | ✓ Transparent pricing, no hidden fees |
| LiteLLM crashes at 500 req/s | ✓ Single binary, no DB in hot path |
Switch models by changing a string. Format conversion, streaming, and tools are handled for you.
Sign up in 30 seconds. No credit card needed for the free tier.
Point your existing OpenAI SDK to api.routers.io. That's it.
OpenAI, Anthropic, DeepSeek, Google, and Qwen through one endpoint.
# Any OpenAI-compatible client works — here's raw curl curl https://api.routers.io/v1/chat/completions \ -H "Authorization: Bearer rtr_your_key_here" \ -H "Content-Type: application/json" \ -d '{"model": "qwen-3.5-9b", "messages": [{"role": "user", "content": "Hello!"}]}'
We publish our prices with no visible markup. Self-hosted models are free — rate-limited but always available.
One key. Transparent pricing. No outages. It just works.