API quickstart

This guide takes you from a new Platform project to a verified API response. The API is OpenAI-compatible, but your key, project, limits, usage, and logs remain managed by Mibyan Platform.

Before you start

You need:

A Mibyan account.
A project in Mibyan Platform.
A project API key. The complete value is shown only once.
A server-side environment where the key can be stored.

Never place an API key in browser JavaScript, a mobile app bundle, a URL, a public repository, or a client-side log. Send requests through your own server.

1. Create a project key

Open Platform → API keys, choose the project, and create a key for the environment you need:

test keys are for development and integration checks.
live keys are for production traffic.

Use the narrowest permissions, model access, endpoint access, and expiry that fit the integration. Copy the secret immediately and save it in your secret manager.

2. Configure the client

The canonical API origin is https://api.mibyanai.com; SDKs should use https://api.mibyanai.com/v1 as their base URL.

export MIBYAN_API_KEY="mbn_test_..."
export MIBYAN_BASE_URL="https://api.mibyanai.com/v1"

3. Send a request

curl "$MIBYAN_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $MIBYAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mibyan-4.1",
    "messages": [
      {"role": "user", "content": "اكتب ملخصاً عملياً عن إطلاق منتج جديد."}
    ]
  }'

The response includes the Mibyan model ID and a request_id correlation value. The same request with the official OpenAI SDKs, pointed at the Mibyan base URL:

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["MIBYAN_API_KEY"],
    base_url="https://api.mibyanai.com/v1",
)
result = client.chat.completions.create(
    model="mibyan-4.1",
    messages=[{"role": "user", "content": "Hello"}],
)
print(result.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.MIBYAN_API_KEY,
  baseURL: "https://api.mibyanai.com/v1",
});
const result = await client.chat.completions.create({
  model: "mibyan-4.1",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(result.choices[0].message.content);

curl "$MIBYAN_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $MIBYAN_API_KEY" \
  -H "Content-Type: application/json" \
  -N \
  -d '{"model":"mibyan-4.1","stream":true,"messages":[{"role":"user","content":"Hello"}]}'

How a request travels

Every response carries x-request-id for support and x-mibyan-api-version (currently 2026-08-11), the stable public boundary version.

4. Verify the integration

Check the following before moving to production:

The request succeeds with the expected project key.
The response model is mibyan-4.1.
The request appears under the same project in Usage and Logs.
Invalid keys return 401 and disallowed endpoints return 403.
Your server does not log the complete key or prompt content unnecessarily.

Common first-request errors

Status	Meaning	What to check
`400`	Invalid request	JSON, messages, token limits, and supported fields
`401`	Authentication failed	Bearer token, key status, expiry, and environment
`403`	Permission denied	Project, endpoint, model, or service-account policy
`404`	Model or endpoint unavailable	Use the exact `/v1` path and the ID returned by `/models`
`429`	Rate or budget limit	Response headers, project policy, and retry timing
`5xx`	Temporary service issue	Retry only according to errors and retries

Guides

Endpoints

Before you start

1. Create a project key

2. Configure the client

3. Send a request

How a request travels

4. Verify the integration

Common first-request errors

​Before you start

​1. Create a project key

​2. Configure the client

​3. Send a request

​How a request travels

​4. Verify the integration

​Common first-request errors

Before you start

1. Create a project key

2. Configure the client

3. Send a request

How a request travels

4. Verify the integration

Common first-request errors