Temu Scraper — Products, Categories & Home Feed

Extract public product data from Temu into a clean, structured Apify dataset: prices, ratings, review counts, images, video, SKU options, specifications and category taxonomy.

No login and no browser required, so runs are fast and inexpensive.

What it scrapes

Surface	Example start URL	What you get
Product detail	https://www.temu.com/-g-601099513224343.html	Full record: price, rating, review count, gallery, video, SKU options, specifications, return policy, breadcrumbs
Category listing	https://www.temu.com/category-o3-248.html	Up to 120 products per category with price, rating, review count, image, sales signal
Home feed	https://www.temu.com/	Temu's recommendation feed, paginates deep (800+ items)

You can also point at a product or category by ID alone — https://www.temu.com/-g-<goodsId>.html or https://www.temu.com/-o3-<optId>.html. No need to find the full product URL first.

What it does not scrape

Being direct about this so you don't buy the wrong tool:

• Keyword search is not supported. Search results cannot be retrieved reliably enough to sell, so search URLs are rejected up front rather than quietly returning products that don't match your query. Browse by category instead — that's the supported way to discover products.
• Store / merchant pages (mall.html) are not supported.
• Review text is not available. Temu does not expose review bodies or author names to this Actor, so each review comes back with its rating and date only, flagged reviewTextRedacted: true. Ratings, review counts and dates are accurate. If you need the written reviews themselves, this Actor cannot supply them.

Input

Field	Type	Default	Description
startUrls	array	https://www.temu.com/	Product, category or homepage URLs. Search URLs are rejected.
maxItems	integer	40	Maximum records to push (1–5000).
getProductDetails	boolean	false	Fetch each listing item's product page for specifications, SKUs and gallery. Costs one extra request per item.
extractReviews	boolean	false	Include the per-product review array (ratings and dates only — see above).

{
  "startUrls": [
    { "url": "https://www.temu.com/category-o3-248.html" },
    { "url": "https://www.temu.com/-g-601099513224343.html" }
  ],
  "maxItems": 100,
  "getProductDetails": false
}

Output

One row per product. Fields appear only when Temu publishes them for that page type, so listing rows stay compact while enriched rows carry more.

Field	Type	Example
goodsId	string	"603014127574739"
title	string	"Wireless Earbuds, Bluetooth 5.4..."
price / priceText	number / string	66.71 / "$66.71"
marketPrice / marketPriceText	number / string	79.99 / "$79.99"
discountPercentage / savingsAmount	number	17 / 13.28
currency	string	"USD"
rating	number	4.9
reviewCount	integer	401
salesTip / salesVolume	string / number	"4.1K+ sold" / 4100
imageUrl / galleryImages	string / array	Temu CDN URLs
hasVideo / videoUrl	boolean / string	product video when present
productUrl	string	canonical https://www.temu.com/-g-<goodsId>.html
categoryId / categoryName / categoryPath / breadcrumbs	string / array	category taxonomy
specifications / skus / returnPolicy	object / array	detail rows only
reviews	array	ratings + dates (reviewTextRedacted)
sourcePageType / sourceUrl / extractedAt	string	provenance

productUrl is always the canonical form, so you can feed output rows straight back in as start URLs.

Proxy

Residential proxy is required and is configured automatically — you don't need to set anything up. Each type of page is routed appropriately on your behalf. Supplying your own proxyConfiguration overrides this and will usually lower success rates, so leave it unset unless you have a specific reason.

Known limitations

• A category can occasionally return no results. Temu does not always serve a given category on demand. The Actor retries automatically before giving up, but a run can still finish with zero rows for that category. Running it again a few minutes later almost always works.
• Some words appear as * in titles and descriptions (for example "with * Display Charging Case"). Temu masks certain brand and claim terms in its own product text; this is how the data is published, not a parsing error.
• Category listings return up to ~120 products each. To cover more, use several category URLs rather than expecting deeper paging within one.
• Every row carries its own currency, so check that field rather than assuming one currency across a dataset.

Notes

The Actor finishes successfully with an empty dataset rather than failing when a page genuinely has no products, so a zero-row run means "nothing available right now", not "broken".