Mercado Libre Scraper & API — 17 Countries

What it does

Mercado Libre is not one marketplace but seventeen, and they do not behave alike: the large ones return thousands of results per query and have catalog product pages, while the small ones return dozens and have no catalog at all. This Actor reads all seventeen with one parser, and puts a currency on every single row because six of them mix the local currency and US dollars on the same results page.

{
  "item_id": "MLV12345678",
  "title": "Apple iPhone 13 128 Gb Medianoche",
  "url": "https://articulo.mercadolibre.com.ve/MLV-12345678",
  "price": 580.8,
  "currency": "USD",
  "currency_symbol": "US$",
  "price_local": 476314,
  "currency_local": "VES",
  "exchange_rate_implied": 820.1,
  "condition": "Nuevo",
  "shipping_text": "Envío gratis",
  "link_kind": "publication",
  "is_catalog": false,
  "position": 3,
  "query": "iphone",
  "site_id": "MLV",
  "scraped_at": "2026-09-13T10:22:41Z",
  "from_cache": false,
  "data_age_hours": 0
}

Why this one

Use cases

• Track prices for a product category across several Latin American marketplaces with one comparable schema.
• Measure how a brand is priced in Venezuela in both dollars and bolívares, using the dual price rather than an external exchange rate.
• Build a competitor catalogue in one country, with seller, condition and shipping text as the marketplace serves them.
• Compare marketplace depth between countries before deciding where a product is worth listing.
• Feed a pricing model rows that each declare their own age through `from_cache`, `scraped_at` and `data_age_hours`.

Input

Every field has a default, and the defaults are deliberately small so a first run is cheap enough to inspect before you commit to a sweep. This table mirrors the Actor's own input schema field for field.

Output and fill rates

A field being in the schema is not the same as it having a value. The percentages below were counted on real runs; the sample sizes are in Measurements. Anything not listed here is not promised.

Every key is always present. A field that exists but is empty comes back as explicit null, so a parser never has to guess.

Datasets

Different record types go to different datasets, so the main table never carries columns that are blank on most rows.

Pricing

Pay per delivered result. Charges are applied as each row is produced rather than in a lump at the end, so an aborted run bills only for what it actually gave you.

Measurements

Each figure is shown with the method that produced it. A benchmark without a method is a marketing claim wearing a number's clothes.

What it will not do

Stated plainly so you can judge fit before spending anything.

• Brazil (MLB) and Cuba (MCU) are refused. Brazil is out until there is a documented LGPD basis; Cuba is unreviewed and its proxy route is unverified.
• Roughly 2,000 items per query is a hard ceiling. Paginating an Argentine `iphone` search to the end reaches about 39% of what the site itself says it has.
• `categorySlicing` is the only route past that ceiling, and how much extra coverage it gives has not been measured and is not promised.
• Rating, reviews, location, stock ranges, attributes, seller id, description and the dual price appear on no marketplace's listing pages. They need `scrapeDetail`, which multiplies page loads by about 40.
• Stock and sales figures are ranges, not numbers: `+25 vendidos` is returned as text, and any derived number is named `_min` because a minimum is what it is.
• Buyer questions and reviews are out of scope entirely — that is text written by identifiable people.
• Seller phone numbers and email addresses are never returned, and seller profile pages under `/perfil/` are never touched.
• Coverage beyond page 1 varies by country and by run. In one pass Colombia and Chile returned the anti-bot wall for both URL forms.
• One snapshot is one moment: cached rows are at most 24 hours old and every row says how old it is.

Privacy

• No account, no session cookies, no OAuth and no API key: the Actor never signs in.
• CAPTCHAs are never solved, by any method, and a sustained CAPTCHA rate stops the run.
• Buyer questions and reviews are not scraped, because that is text written by identifiable people.
• No seller phone numbers and no email addresses, which some Actors in this category do return.
• Seller profile pages under `/perfil/` are disallowed by robots.txt and are never requested.
• Images are not rehosted; rows carry image URLs.
• Listings are cached for at most 24 hours and every row declares its age.
• Removal requests, quoting the `item_id`: privacy@actorstack.dev

See also the data removal process.

Guides for this Actor

• How to scrape Mercado LibreA working method for extracting listings and product pages from any of the 17 Mercado Libre marketplaces, including the two decisions — pagination and detail pages — that decide what a run costs.
• API alternativeThe official Mercado Libre API requires OAuth and returns 403 to an anonymous request. A comparison of what the API gives an authorised caller, what scraping gives anyone, and which fields exist in only one of the two.
• The 17 marketplacesA reference table of every Mercado Libre site id, its country, its currency, the depth of a single query and whether it has catalog pages, mixed currencies and installments.
• Mixed-currency pagesIn Uruguay, Paraguay, the Dominican Republic, Nicaragua, Guatemala and Panama the local currency and US dollars appear in the same list of results — which makes a per-country currency column quietly wrong.
• Venezuela's dual priceVenezuelan product pages show a dollar price and a bolívar price at once. Deriving the implied rate from the page's structured data rather than the rendered price is the difference between a stable number and one that drifts.
• Pagination and robots.txtMeasured in September 2026: the plain paginated path returns the anti-bot wall and the `_NoIndex_True` form returns results — and robots.txt disallows the second by name. There is currently no polite pagination available.
• item_id and link shapesMercado Libre results mix `articulo.…/MLA-…`, `/p/MLA…` and `/up/MLAU…` links. Deriving the item id from the URL looks fine until the third shape appears, which can be nearly half a page.
• The anti-bot wallMercado Libre's anti-bot response is a well-formed page with a 200 status and zero listings. Six different responses have to be told apart, and only one of them is data.
• Stock and sales rangesThe site shows `+25 vendidos` and the official API answers `RANGO_1_50`. Any dataset with an exact sales integer in it invented that integer.
• Listing versus detailWhich fields only exist on product pages, what turning on `scrapeDetail` does to a run's page loads and its bill, and how to get the enriched fields for only the rows that need them.