Support &
Resources
Getting Started
How do I deploy Aurora on my local infrastructure?
Aurora is distributed as a single static binary. You can deploy it using Docker, Kubernetes, or as a native Go service. The fastest way to get started is using our official Docker image.
docker run -d aurora/gateway:latestEnsure you have mapped your config.yaml to the container volume for proper provider routing. Refer to the Full Deployment Guide for kubernetes specs.Can I use my existing OpenAI/Anthropic API keys?
Yes. Aurora acts as a transparent proxy that accepts OpenAI- and Anthropic-compatible request formats. Simply configure your existing API keys in the provider section of config.yaml and Aurora will route requests accordingly.API keys are encrypted at rest and never logged. You can rotate them at any time via the admin dashboard or the CLI without restarting the gateway.
What is the latency overhead of routing through Aurora?
Aurora adds under 2ms of p99 latency per routing decision. Our benchmarks show that intelligent routing (including fallback, caching, and rate limiting) typically improves overall response times by 15–40% compared to direct provider calls.
The gateway is written in Go and utilizes zero-copy buffer management for maximum throughput.
How does the intelligent fallback mechanism work?
Aurora monitors provider health in real-time. If a primary provider returns a 5xx error, times out, or exceeds a configurable latency threshold, the request is automatically retried on the next healthy provider in your priority list.
You can configure retry count, timeout windows, and circuit-breaker thresholds per provider group. All fallback events are logged for observability.
Does Aurora support streaming responses?
Yes. Aurora fully supports Server-Sent Events (SSE) streaming for all major providers including OpenAI, Anthropic, and Google. Streaming is handled with zero-copy forwarding and does not buffer response payloads.
Token-level streaming metrics are available in real-time through the admin dashboard.
Can I restrict which models are available to my users?
Absolutely. Aurora supports granular access controls through model allow-lists and deny-lists. You can create multiple user roles with different model access scopes, rate limits, and credit allocations.
This is especially useful for multi-tenant deployments where different teams require access to different provider models.
Can't find what you're looking for?
Our community and documentation are updated daily. If your technical challenge is unique, our engineering team is standing by to help.
Need Direct Support?
Our technical engineers are available for architecture consultation.
Contact Engineer