# Quickstart

> Create a key and send a first request in a few minutes.

## Create a key

Open [API keys](https://deference.si/keys) and select **Create key**. Name it after what will use it, and copy the key when it appears: it is shown once.

New accounts get up to $2.50 of [free credit](https://deference.si/docs/concepts/free-credit) for open-weight text models, like the one below. For any other model, [add credit](https://deference.si/overview) with USDC or INFERENCE.

## Set the key

**macOS and Linux**

```bash
export DEFERENCE_API_KEY="sk-df-..."
```

**Windows**

```powershell
$env:DEFERENCE_API_KEY = "sk-df-..."
```

## Send a request

**curl**

```bash
curl https://deference.si/v1/chat/completions \
  -H "Authorization: Bearer $DEFERENCE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v3.2",
    "messages": [{ "role": "user", "content": "Say hello in one sentence." }],
    "max_tokens": 100
  }'
```

**Python**

```python
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://deference.si/v1",
    api_key=os.environ["DEFERENCE_API_KEY"],
)

response = client.chat.completions.create(
    model="deepseek/deepseek-v3.2",
    messages=[{"role": "user", "content": "Say hello in one sentence."}],
    max_tokens=100,
)
print(response.choices[0].message.content)
```

**TypeScript**

```typescript
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://deference.si/v1",
  apiKey: process.env.DEFERENCE_API_KEY,
});

const response = await client.chat.completions.create({
  model: "deepseek/deepseek-v3.2",
  messages: [{ role: "user", content: "Say hello in one sentence." }],
  max_tokens: 100,
});
console.log(response.choices[0].message.content);
```

Swap `model` for any id from [Models](https://deference.si/models). Requests that do not meet the [free-credit rules](https://deference.si/docs/concepts/free-credit#what-it-pays-for) need credit.

## Check it in Activity

Open [Activity](https://deference.si/activity). The request is the top row, with its model, tokens and cost. Until the charge settles, the cost is marked as an estimate, then shows the final amount.

Deference rejects invalid keys, missing models, insufficient credit, and its own rate limits before opening an Activity row. Provider errors such as 404 and 429 do appear. A provider 429 counts as a failure.

**If it fails**

| You see                        | Cause                                                                        | Fix                                                                                                                                                                       |
| ------------------------------ | ---------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `401 missing_api_key`          | The variable is empty, so no key was sent                                    | Set the variable in the shell that runs the code                                                                                                                          |
| `401 invalid_api_key`          | The key is mistyped                                                          | Set the variable again from your saved key, or create a replacement. Keep the value out of logs                                                                           |
| `402 insufficient_credit`      | The account has no available credit                                          | [Add credit](https://deference.si/overview)                                                                                                                                                   |
| `402 free_credit_not_eligible` | The account has only free credit, and this model or request does not qualify | Check the [model price and request restrictions](https://deference.si/docs/concepts/free-credit#what-it-pays-for). Lower `max_tokens` if the worst-case estimate exceeds $0.05, or add credit |
| `404 model_not_found`          | The model id is not in the catalog                                           | Copy an id from the Models page                                                                                                                                           |
| `429 rate_limited`             | Too many requests in a minute                                                | Wait for the `retry-after` seconds                                                                                                                                        |

## Set up a tool

Point your coding tool at Deference next. Every guide is in [Coding tools](https://deference.si/docs/coding-tools).

* [Claude Code](https://deference.si/docs/coding-tools/claude-code) (Messages)
* [Codex](https://deference.si/docs/coding-tools/codex-cli) (Responses)
* [Cursor](https://deference.si/docs/coding-tools/cursor) (Chat Completions, beta)

Writing code instead? See [what your key can do](https://deference.si/docs/what-you-can-do).