About ProxyStacks Relay
ProxyStacks and Relay
Relay is a multi-model API gateway built and operated by ProxyStacks: one key and one endpoint for the models configured in your console, metered with prepaid balance.
- Protocol compatible
- One endpoint accepts OpenAI-compatible Chat Completions and Responses requests as well as Anthropic Messages, so existing SDKs keep working.
- Keys and routes
- Issue and revoke Relay keys in the console; the server stores only the SHA-256 hash of each key. Model routes are configured centrally, and clients pass only a model ID.
- Metering and balance
- Prepaid balance are deducted per request at the published token rates for the selected model, settling after a streaming response ends. Balances, usage and request IDs stay traceable.
- Developer tooling
- Claude Code, Codex and opencode can point at Relay directly: swap the base URL, key and model ID as shown in the docs.
Operation and boundaries
Backend capacity is procured from paid commercial cloud and AI API channels. We do not represent any foundation model vendor or sell upstream vendor accounts. Payments are processed by Waffo Pancake as merchant of record, including invoices and applicable indirect taxes.
Start with your first API request.
Create a key, point your existing client at Relay, and start calling with prepaid balance.