# Deepseek Autoparse

Source: https://docs.scrappey.com/docs/deepseek-autoparse

> Use the DeepSeek large language model to convert dynamic web content into structured JSON in a single request.

## Command

`"Deepseek AI Autoparse"` — costs **2 credits** per successful request.

- **Model**: `deepseek`
- **Autoparse**: Set `"autoparse": true`
- **Optional**: Supply `"structure"` and `"cssSelector"` to guide parsing.

## Parameters

| Parameter     | Type    | Required | Description                                                                 |
| ------------- | ------- | -------- | --------------------------------------------------------------------------- |
| `model`       | string  | Yes      | Must be `"deepseek"`                                                        |
| `autoparse`   | boolean | Yes      | Enable autoparsing                                                          |
| `cssSelector` | string  | No       | Specific region of the website to parse                                     |
| `structure`   | object  | No       | Structure of the desired output. If omitted, model infers one automatically |

## Example

```bash
curl -X POST "https://publisher.scrappey.com/api/v1?key=YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "cmd": "request.get",
    "url": "https://www.imdb.com/title/tt0087363/?ref_=ls_t_10",
    "filter": [
      "autoparse"
    ],
    "model": "deepseek",
    "autoparse": true,
    "cssSelector": "section[class='ipc-page-background ipc-page-background--base sc-afa4bed1-0 iMxoKo']",
    "structure": {
      "movie_info": {
        "title": "str",
        "language": "str"
      },
      "duration": "str",
      "release_date": "str",
      "rating": {
        "average_rating": "float",
        "total_votes": "int"
      },
      "certification": "str",
      "cast": [
        {
          "actor_name": "str",
          "role": "str"
        }
      ],
      "crew": [
        {
          "crew_name": "str",
          "role": "str"
        }
      ]
    }
  }'
```

Example response:

```json
{
  "solution": {
    "verified": true,
    "type": "browser",
    "autoparse": {
      "movie_info": {
        "title": "Gremlins",
        "language": "English"
      },
      "duration": "1h 46m",
      "release_date": "June 8, 1984",
      "rating": {
        "average_rating": 7.3,
        "total_votes": 259000
      },
      "certification": "PG",
      "cast": [
        {
          "actor_name": "Zach Galligan",
          "role": "Billy"
        },
        {
          "actor_name": "Phoebe Cates",
          "role": "Kate"
        }
      ],
      "crew": [
        {
          "crew_name": "Joe Dante",
          "role": "Director"
        },
        {
          "crew_name": "Chris Columbus",
          "role": "Writer"
        }
      ]
    }
  },
  "timeElapsed": 41380,
  "data": "success",
  "session": "b0380e15-3381-413c-8cad-daefe574afb1"
}
```

## Notes

DeepSeek uses a Mixture of Experts (MoE) architecture for fast inference. It is open-source, multilingual, and performs well on coding, math, and structured parsing tasks. DeepSeek-V3 access is available for free.