Dongchedi Scraper: China Car Specs, Dealer Prices & Owner Reviews (懂车帝)

Get structured data from Dongchedi (懂车帝), one of China's largest car websites. The Actor covers the full new-car catalog of 4,700+ series. For every trim (车型) you get the full spec sheet of about 290 fields, the local dealer price and discount in the city you choose, and owner reviews (口碑) with scores for each category.

• One Actor for specs, prices and reviews. You get the parameter table (参数配置), the dealer quote (经销商报价) and owner reviews for each trim in one run.
• No login and no cookies. Only public data is collected, so your account is never involved.
• Low price. $3 per 1,000 trims with full specs and prices, and $1 per 1,000 reviews.

What you can scrape

Mode	What you get
series_detail (default)	All trims of a series with full specs and dealer prices, plus owner reviews
specs	Trims with full specs and prices only. Takes series ids or trim ids
reviews	Owner reviews only: overall score, category scores, purchase price, city, text and photos
series_list	The car catalog: all series, or the series of one brand, with price range, Dcar score and trim ids
search	Dongchedi content search: owner posts and articles for a keyword

Input

Field	Description
mode	One of the modes above
seriesIds	Series URLs such as https://www.dongchedi.com/auto/series/20154, or plain ids
brandIds	Brand ids, for example 2 for Audi (奥迪) or 16 for BYD (比亚迪). Use them in series_list, or to scrape a whole brand
carIds	Trim ids or URLs (.../model-259371). Used in specs mode
cityName	City used for dealer prices, written in Chinese. Default 北京
includeSpecs, includeDealerPrices	Turn the spec sheet or the dealer prices on or off
onlyOnSale	Keep only trims that are on sale (default). Turn it off to also get discontinued and upcoming trims
maxCarsPerSeries, maxReviewsPerSeries, maxSeries	Limits. Set maxReviewsPerSeries: 0 to skip reviews
keywords, maxResultsPerKeyword	Keywords and result limit for search mode

{
  "mode": "series_detail",
  "seriesIds": ["https://www.dongchedi.com/auto/series/20154"],
  "cityName": "上海",
  "maxReviewsPerSeries": 100
}

To scrape the whole catalog of one brand:

{"mode": "specs", "brandIds": ["16"], "maxSeries": 50}

Output

Every dataset item has a type field. It is one of car, review, series, post or error.

{
  "type": "car",
  "series_id": 20154, "series_name": "星愿", "brand_id": 858, "brand_name": "吉利银河",
  "car_id": 259371, "car_name": "310km 向往版", "year": 2026,
  "official_price": "6.48万", "dealer_price": "6.18万", "cut_price": "0.30万", "owner_paid_price": "6.14万",
  "city": "北京", "sale_status": "on_sale", "dcar_score": 3.42, "body_type": "轿车",
  "url": "https://www.dongchedi.com/auto/series/20154/model-259371",
  "specs": {
    "官方指导价": "6.48万", "厂商": "吉利银河", "级别": "小型车", "能源类型": "纯电动",
    "纯电续航里程(km)CLTC": "310", "最大功率(kW)": "58kW(79Ps)", "长x宽x高(mm)": "4135x1805x1570",
    "轴距(mm)": "2650", "整备质量(kg)": "1215", "整车保修期限": "4年或15万公里",
    "驾驶辅助影像": "倒车影像", "最小离地间隙(mm)": null, "...": "about 290 fields in total"
  }
}

In specs, the value 标配 means the feature is standard, 选装 means it is optional, and null means the trim does not have it. Prices are in the units the site uses: 万 means 10,000 CNY.

{
  "type": "review", "review_id": "7669767269450256408", "series_id": 20154, "car_name": "480km 探索版",
  "score": 4.5,
  "scores": {"appearance": 5.0, "interior": 5.0, "space": 5.0, "configuration": 4.0, "range": 4.0, "driving": 4.0},
  "content": "这个价位我感觉够开了，配置够用，底盘配置给的也够，就是没有座椅通风按摩",
  "bought_time": "2026-07", "location": "常州", "purchase_price": "8.23万", "range": "410.00km",
  "digg_count": 45, "comment_count": 19, "create_time": "2026-08-03T11:36:00+00:00",
  "author_name": "皮蛋瘦肉周ha", "author_hash": "a669a92593b11e35"
}

Review scores are on a 5-point scale. User ids are never returned. Instead each review carries a short author_hash, which lets you group reviews by the same author.

{
  "type": "series", "series_id": 99, "series_name": "奥迪A6L", "brand_name": "奥迪",
  "official_price": "32.29万", "dealer_price": "26.33-45.67万", "dcar_score": 4.12,
  "car_ids": [249016, 249017, 249018], "car_count": 10, "url": "https://www.dongchedi.com/auto/series/99"
}

Use cases

• Competitive analysis and pricing for car makers and dealers. Compare trims, equipment and local discounts across brands.
• Spec databases and comparison tools. Load the complete Chinese car catalog with about 290 parameters per trim.
• EV market research. Get CLTC range, battery, charging and smart-cockpit specs for Chinese EVs.
• Voice of the customer. Analyse what owners praise or complain about, with an LLM or with sentiment analysis, using category scores and the price each owner paid.

Use it from code, n8n or an AI agent

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("xiaoliang527/dongchedi-car-data").call(
    run_input={"seriesIds": ["20154"], "cityName": "北京", "maxReviewsPerSeries": 30})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    if item["type"] == "car":
        print(item["car_name"], item["dealer_price"], item["specs"].get("纯电续航里程(km)CLTC"))

MCP (Claude, Cursor): connect to https://mcp.apify.com?tools=xiaoliang527/dongchedi-car-data and send your Apify token as a Bearer header.

Pricing

Event	Price
Car trim, with full specs and dealer price	$0.003 ($3 / 1,000)
Owner review	$0.001 ($1 / 1,000)
Series from the catalog list	$0.001
Search result (post)	$0.001

Errors such as an invalid URL or an unknown series are returned with type: "error" and are not charged. You can set a maximum cost per run in the run options. When the Actor reaches it, it stops cleanly.

Limits

• Not every trim has a dealer quote. When there is none, dealer_price is null (the site shows "暂无报价").
• Review items contain the review text, photos and scores. Comment threads under reviews are not expanded.
• Image URLs are signed by the site and expire after some time. Download the images soon after the run if you need them.
• Used cars, sales rankings and posts in owner communities (车友圈) are not included.
• Requests are rate-limited to about 2–3 per second to stay polite. A trim with full specs takes about 0.5 s.
• Only public data is collected. Nothing behind login is accessed.