Advanced Features
Auto Failover
When your primary model provider experiences downtime, OneRouter automatically routes traffic to your backup model — with zero dropped requests and zero manual intervention.
How It Works
OneRouter monitors upstream provider health in real-time. When a provider returns repeated 5xx errors or exceeds a timeout threshold, traffic is automatically redirected to the next model in your failover chain. Once the primary recovers, traffic resumes normal routing.
The failover chain is configured per-model in the Dashboard — no custom headers or code changes required on the client side.
Configuring Failover in the Dashboard
- Log in to the OneRouter Dashboard at api.onerouter.app/dashboard
- Navigate to Channels in the sidebar
- Select the model channel you want to protect with failover
- In the channel settings, locate the Model Mapping or Priority configuration
- Add your backup model(s) in priority order — the first available model in the list will serve traffic
- Set the timeout and retry count thresholds that trigger a failover
- Save the channel configuration
Building a Resilient Failover Chain
For production workloads, we recommend a cross-provider failover strategy:
| Priority | Model | Provider | Role |
|---|---|---|---|
| 1 (Primary) | GPT-4o | OpenAI | Default — best overall quality |
| 2 (Fallback) | Claude Sonnet 4.6 | Anthropic | Activates if OpenAI is down |
| 3 (Last resort) | Gemini 2.5 Flash | Budget-safe safety net |
Pro tip: Always include at least one model from a different provider in your failover chain. Provider-level outages are the most common failure mode. A cross-provider chain (OpenAI → Anthropic → Google) provides the strongest protection.
Monitoring Failover Events
You can track failover activity through:
- Dashboard logs — Filter by channel to see failover-triggered requests
- Webhooks — Subscribe to
system.incidentevents for real-time failover notifications (see Webhooks) - Usage analytics — Per-model usage spikes indicate failover activity