mojaq-openai

A drop-in replacement for the openai client that routes every request through your MOJAQ AI gateway — so you get EU-hosted usage/cost logging, budget caps, a kill-switch and automatic model fallback with one import and no other code changes.

npm i mojaq-openai openai

import OpenAI from 'mojaq-openai';

const client = new OpenAI(); // reads MOJAQ_API_KEY from the environment

const res = await client.chat.completions.create({
  model: 'gpt-4o-mini',
  messages: [{ role: 'user', content: 'Hello' }],
});

That's the whole change — everything else is the standard OpenAI SDK (streaming, tools, embeddings, …), because mojaq-openai extends it.

Setup

1. Create a MOJAQ account and add a provider in AI Gateway (any OpenAI-compatible provider — OpenAI, OpenRouter, Groq, …). Your provider key is stored encrypted in MOJAQ and injected server-side; it never ships in your app.
2. Set MOJAQ_API_KEY to your MOJAQ API key.
3. Swap import OpenAI from 'openai' → import OpenAI from 'mojaq-openai'.

| Env var | Default | Purpose | | --- | --- | --- | | MOJAQ_API_KEY | — | Your MOJAQ key (auth to the gateway) | | MOJAQ_BASE_URL | https://api.mojaq.com/ai/v1 | Override if self-hosting |

What you get

• Usage, cost and latency for every call, per model.
• Budget caps + kill-switch to stop runaway spend instantly.
• Model fallback if a provider errors.
• EU-hosted request path.

MIT licensed. Made by MOJAQ.