Drop back to raw OpenAI

The wire is iso-OpenAI, so you can skip the SDK and point the official OpenAI client at Fleeex.

@fleeex/sdk is never a lock-in. The wire is iso-OpenAI, so you can skip the package entirely and point the official OpenAI SDK at the Fleeex base URL. Every request is byte-identical to what @fleeex/sdk sends, and there's a non-regression test asserting exactly that.

import OpenAI from "openai";
 
const client = new OpenAI({
  apiKey: process.env.FLEEEX_API_KEY!,
  baseURL: "https://api.fleeex.dev/v1", // note the trailing /v1
  defaultHeaders: { "x-fleeex-user": "user-123" },
});
 
const completion = await client.chat.completions.create({
  model: "nova-lite",
  messages: [{ role: "user", content: "Hello!" }],
});

What you give up

Compared to @fleeex/sdk, going raw means:

• You manage the connection. baseURL must include /v1, and you set x-fleeex-user yourself on every request.

• 402 is harder to use. It arrives as OpenAI.APIError (status === 402). The top-up link is in the response body, but the OpenAI SDK surfaces only the body's error field, so topupUrl isn't readily accessible. @fleeex/sdk lifts it into PaymentRequiredError.topupUrl for you.

• No billing helpers. getBalance(), getConnection(), and getUsageSummary() aren't part of the OpenAI SDK, so call the routes yourself, with the same bearer key and x-fleeex-user. They sit at the root, not under /v1:

  Helper	Route
  getBalance()	GET /balance
  getConnection()	POST /connect, where redirectUri travels in the body so it stays out of access logs
  getUsageSummary()	GET /usage/summary

You give up nothing on the model surface: models.list() and models.retrieve() are plain OpenAI calls that already go through this client. See Choosing a model.

base_url at a glance

Everything else (request bodies, streaming, response shapes) is identical.