R9S, also called Route Tokens, is a distributed AI gateway for production model traffic. It gives applications one stable access layer for multiple models, providers, regions, and backend suppliers.
The core API endpoint is:
https://gw.r9s.ai/v1Applications send OpenAI-compatible, Anthropic-compatible, or SDK requests to R9S. The gateway authenticates the call, chooses a healthy upstream route, forwards the request, and records usage and telemetry for operations and cost control.
What R9S solves
| Need | R9S capability | Result |
|---|---|---|
| Unified model access | One endpoint and one authentication pattern | Lower integration and migration cost |
| Multi-provider routing | Model and supplier aggregation | More choice and better continuity |
| Low-latency access | Global edge nodes and region-aware routing | Shorter request paths for cross-region workloads |
| Reliability | Explicit fallback chains and provider failover | Fewer model-provider incidents exposed to users |
| Observability | Metrics, logs, traces, and request correlation | Faster debugging and capacity decisions |
| Governance | API keys, quotas, and billing | Clear ownership of usage and spend |
Request path
A typical request moves through these layers:
- Your application calls the R9S gateway.
- R9S validates the API key and request format.
- The gateway evaluates model availability, region, provider health, capacity, and fallback policy.
- R9S forwards the request to the selected upstream path.
- The response is returned in the expected compatible format.
- Usage, latency, cost, and route metadata are recorded for monitoring.
Documentation map
| Section | Start here when you need to |
|---|---|
| Quick Start | Make the first authenticated model request |
| Core Concepts | Understand gateway routing, fallback, attribution, and keys |
| API Reference | Find base URLs, headers, endpoint families, and response conventions |
| Monitoring and Operations | Track production health, cost, logs, and rollout behavior |
| Agent Integrations | Configure coding agents to use R9S as their model gateway |
Gateway values
R9S is designed around a small set of operating values:
- Keep application integration simple.
- Route across providers and regions without changing application code.
- Make fallback behavior explicit and observable.
- Preserve streaming and non-streaming compatibility.
- Attribute usage to applications, teams, and organizations.
Next step
Continue with Quick Start to verify your key and send the first request.