ARouter helps developers source, route, and observe AI usage across any provider. We believe the future is multi-model and multi-provider.Documentation Index
Fetch the complete documentation index at: https://docs.arouter.ai/llms.txt
Use this file to discover all available pages before exploring further.
Why ARouter?
Price and Performance. ARouter continuously tracks pricing, latency, and throughput across all upstream providers, so you always get the best deal without changing a line of code. Standardized API. One endpoint, one API key — whether you’re calling OpenAI, Anthropic, Gemini, DeepSeek, or any other provider. You can switch models or providers by changing a single string. Consolidated Billing. A single credit balance covers all providers. No more juggling invoices from a dozen vendors. Higher Availability. ARouter monitors provider health in real time and routes around outages automatically. You can also supply ordered candidate models so requests always find a path to completion. Higher Rate Limits. By aggregating traffic across providers and regions, ARouter enables throughput that would be impractical to achieve on a single provider account. Unified Observability. Every request is metered — tokens, latency, cost, status — visible in your dashboard and exportable for accounting. Scoped API keys, spending limits, and per-model restrictions let you run multi-team deployments safely.Design Principles
Drop-in Compatible
ARouter’s API is a strict superset of the OpenAI Chat Completions API. Any SDK or tool already using that interface — official OpenAI SDKs, LangChain, Vercel AI SDK, and more — works without changes beyondbase_url and api_key. Native Anthropic and Gemini SDK paths are also supported.