@axemere/gateway-openai

Drop-in OpenAI client factory for the Axemere AI Gateway.

Returns a standard OpenAI instance pre-configured to route through the gateway. All existing OpenAI code works unchanged — the gateway adds cost controls, policy enforcement, and an append-only audit ledger to every call.

Install

npm install @axemere/gateway-openai

Usage

import { openaiClient } from "@axemere/gateway-openai";

const client = openaiClient(); // reads AXEMERE_GATEWAY_URL + AXEMERE_GATEWAY_TOKEN

const response = await client.chat.completions.create({
  model: "gpt-4o-mini",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(response.choices[0].message.content);

Streaming works exactly as it does with the standard openai package.

Configuration

| Env var | Description | |---------|-------------| | AXEMERE_GATEWAY_URL | Gateway base URL, e.g. http://localhost:7080 | | AXEMERE_GATEWAY_TOKEN | Gateway token issued by the gateway |

Links

• Axemere AI Gateway
• Documentation
• GitHub

License

MIT