> ## Documentation Index
> Fetch the complete documentation index at: https://docs.heyaskr.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# DeepSeek

> DeepSeek: open-weight reasoning and code at a low price.

DeepSeek is an open-weight family that punches well above its price, especially
for reasoning and code. A strong default when you want capable output without
paying frontier prices.

## Best for

* High-volume workloads where cost matters.
* Reasoning and code at a low price point.
* Self-serve experimentation and prototyping.

|                      |                                 |
| -------------------- | ------------------------------- |
| **Modality**         | Chat & code (text in, text out) |
| **Example model id** | `deepseek-v3`                   |
| **Endpoint**         | `POST /chat/completions`        |

## Example

<CodeGroup>
  ```bash cURL theme={null}
  curl https://api.heyaskr.ai/chat/completions \
    -H "Authorization: Bearer ASKR_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "model": "deepseek-v3",
      "messages": [
        {"role": "user", "content": "Write a Python function to debounce calls."}
      ]
    }'
  ```

  ```python Python theme={null}
  from openai import OpenAI
  client = OpenAI(base_url="https://api.heyaskr.ai", api_key="ASKR_API_KEY")

  resp = client.chat.completions.create(
      model="deepseek-v3",
      messages=[
          {"role": "user", "content": "Write a Python function to debounce calls."},
      ],
  )
  print(resp.choices[0].message.content)
  ```
</CodeGroup>

## Pricing

Priced per token, input and output separately. As a rough guide, expect around
**\$0.30 per 1M input tokens** and **\$1.20 per 1M output tokens**, among the
lowest-cost capable models on Askr.

<div className="askr-img-placeholder">
  <span><strong>Image</strong><br />DeepSeek on Askr: model card and pricing</span>
</div>

<Note>
  The model id and prices above are examples. Confirm the exact id and current
  price against the live [`/v1/models`](/api-reference/list-models) list before you
  rely on them.
</Note>