Getting Started
Routing
Plurence routes every request to the right provider based on your model selection, policies, and fallback rules.
How routing works
When a request arrives at the gateway, Plurence resolves the target provider from the model name you specified. Supported providers include OpenAI, Anthropic, and Google.
The model name in your request directly controls which provider receives the call:
| Model name | Provider |
|---|---|
| gpt-4o | OpenAI |
| gpt-4o-mini | OpenAI |
| claude-3-5-sonnet-20241022 | Anthropic |
| claude-3-haiku-20240307 | Anthropic |
| gemini-1.5-pro | |
| gemini-1.5-flash |
Provider credentials
Plurence holds provider API keys on your behalf. Configure them in Settings → Providers. Your traffic is authenticated to the upstream provider using those credentials — your callers only ever see your Plurence API key.
Fallbacks Coming soon
Configure a priority-ordered list of providers for a given model family. If the primary provider returns a 5xx or rate-limit error, Plurence automatically retries on the next provider in the list with no changes required in your application code.
Load balancing Coming soon
Distribute requests across multiple provider accounts or regions using weighted round-robin. Useful for staying within per-organization rate limits at high throughput.