polymer-pay-llm

LLM request routing to the cheapest provider is handled automatically by Polymer Pay's category-based proxy. You specify a model category in the chat completions endpoint, and the service selects the lowest-cost available provider without requiring manual provider selection.

Yes, OpenAI-compatible chat completions are supported through the Polymer Pay proxy using a single API key. You send standardized requests to the proxy endpoint, and it manages all provider authentication, payment negotiation, and failover internally.

When an LLM provider fails during a request, automatic failover redirects the chat completion to the next available provider. The proxy handles retries and provider switching transparently, ensuring your request completes without manual intervention.

Model discovery and per-model pricing are available through dedicated endpoints in the proxy. You can query these to compare available LLM options, view costs, and plan your routing strategy before sending chat completion requests.

Multi-provider LLM routing works with OpenAI-compatible client libraries by pointing them to the Polymer Pay proxy endpoint. The proxy accepts standard /v1/chat/completions requests, so existing code requires only a base URL and API key change.

Using a proxy for LLM failover means all requests route through the Polymer Pay endpoint, requiring network access to the service. Routing is category-based, so you cannot manually select a specific underlying provider, only the cheapest available in a category.