> ## Documentation Index
> Fetch the complete documentation index at: https://docs.llm.kiwi/llms.txt
> Use this file to discover all available pages before exploring further.
# Quickstart
> Step-by-step quickstart for LLM.kiwi: sign up for a free API key, store it safely in an environment variable, and send your first chat completion request.
This guide takes you from a new account to a working API call in a few minutes. No previous API experience is required.
Brand new to programming? Start with the cURL example — it works from any terminal with zero setup.
## Before you start
You need:
* A free LLM.kiwi account
* An internet connection
* One of these options: a terminal, Python 3, or Node.js
No credit card, Docker, local AI model, or server is required.
## Step 1: Create an API key
Open [llm.kiwi/login](https://llm.kiwi/login) and sign in with Google or GitHub.
Go to [llm.kiwi/dashboard](https://llm.kiwi/dashboard), where you can create and manage API keys.
Select **Create key**, give it a recognizable name such as `first-test`, and copy the value beginning with `sk_kiwi_`.
Treat an API key like a password. Save it in a password manager or environment variable, never publish it in GitHub, and revoke it immediately if it leaks.
## Step 2: Make your first request
Choose one option below. Replace `sk_kiwi_...` with your real key only for this first local test.
Open Terminal on macOS or Linux, or PowerShell on Windows, and run:
```bash theme={null}
curl "https://api.llm.kiwi/v1/chat/completions" \
-H "Authorization: Bearer sk_kiwi_..." \
-H "Content-Type: application/json" \
-d '{
"model": "auto",
"messages": [
{"role": "user", "content": "Reply with exactly: Hello from LLM.kiwi!"}
]
}'
```
Recent versions of Windows, macOS, and most Linux distributions include cURL. If `curl` is not recognized, use the Python or JavaScript tab.
Install the OpenAI package:
```bash theme={null}
python -m pip install openai
```
Create `test.py`:
```python theme={null}
from openai import OpenAI
client = OpenAI(
base_url="https://api.llm.kiwi/v1",
api_key="sk_kiwi_...",
)
response = client.chat.completions.create(
model="auto",
messages=[
{"role": "user", "content": "Reply with exactly: Hello from LLM.kiwi!"}
],
)
print(response.choices[0].message.content)
```
Run it:
```bash theme={null}
python test.py
```
Install Node.js 18 or newer, create a new folder, and run:
```bash theme={null}
npm init -y
npm install openai
```
Create `test.mjs`:
```javascript theme={null}
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.llm.kiwi/v1",
apiKey: "sk_kiwi_...",
});
const response = await client.chat.completions.create({
model: "auto",
messages: [
{ role: "user", content: "Reply with exactly: Hello from LLM.kiwi!" },
],
});
console.log(response.choices[0].message.content);
```
Run it:
```bash theme={null}
node test.mjs
```
A successful request prints `Hello from LLM.kiwi!`. The raw API response also contains a `choices` array, with the answer at `choices[0].message.content`.
## Step 3: Move your key out of the code
Hardcoding a key is acceptable only for a quick local test. Before you build an app, store it in an environment variable.
```bash theme={null}
export LLM_KIWI_API_KEY="sk_kiwi_..."
```
```powershell theme={null}
$env:LLM_KIWI_API_KEY="sk_kiwi_..."
```
Then read it in your code:
```python theme={null}
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.llm.kiwi/v1",
api_key=os.environ["LLM_KIWI_API_KEY"],
)
```
```javascript theme={null}
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.llm.kiwi/v1",
apiKey: process.env.LLM_KIWI_API_KEY,
});
```
## What just happened?
1. Your program sent an HTTPS `POST` request to `/v1/chat/completions`.
2. The `Authorization` header identified your account using the API key.
3. `model: "auto"` asked LLM.kiwi to select an available model.
4. The `messages` array supplied the conversation.
5. The API returned the assistant's answer as JSON.
Read the concepts guide for plain-English explanations of endpoints, headers, JSON, models, messages, and tokens.
## If it did not work
| What you see | Most likely cause | Fix |
| ----------------------- | --------------------------------------- | ----------------------------------------------- |
| `401 Unauthorized` | Missing, invalid, or revoked key | Copy the key again and keep `Bearer ` before it |
| `400 Bad Request` | Invalid JSON or missing field | Check commas, quotes, `model`, and `messages` |
| `429 Too Many Requests` | Rate limit reached | Wait briefly and retry with backoff |
| Command not found | cURL, Python, or Node.js is unavailable | Install it or choose another tab |
For detailed fixes, open [Troubleshooting](/troubleshooting).
## Next steps
Learn the few concepts used in every request.
Understand when to use `auto` or `hrllm`.
Add conversations, system instructions, and streaming.
See every request field and response property.