1688 Supplier Finder: Find Suppliers in China, AI-Vetted

New trusted 1688 factories for your product — costed and translated, in your inbox every week.

⚡ First 25 products free · 💵 $0.006 per qualified product · 🤖 ≈ $0.021 all in with AI, on plans with model access or your key · ⏱️ demo in seconds

Try it in 30 seconds. Click Try it — the form is pre-filled with 6 fictional sample listings (their links open a real 1688 search). Listings that fail your filters cost nothing here. → What a run costs · What the AI adds · Run it weekly

⚡ At a glance

What you get	1688 products from trusted suppliers that fit your quantity and margin — English title, supplier score, landed cost, a supplier message
You provide	What you source, English or Chinese (硅胶手机壳), your order quantity and sell price
Output	a ranked sheet (Excel, CSV, Sheets) and an e-mail digest of what is new
Typical run	1 search × 100 listings → ≈ 40 qualified (40 %) in ≈ 2 minutes (estimate); demo: 6 → 4 in 0.5 s
Cost of that run	$0.44 rules only · $0.84 with AI, all-in (the $0.20 scan on your account included); the first 25 are free
Keys / setup	With Apify model access: no key · Other plans: up to 5 live AI verdicts on us, then your own key — a free Gemini / Groq key gives 1–3 AI verdicts per run, a paid one every product
Which plan	Free demo and your own rows (list mode, template): any plan · live search: a plan that runs Store Actors
EU / UK sellers	Landed cost and margin in EUR or GBP, import VAT next to duty — the "EU / UK Amazon seller" preset
Agencies	Your whole roster in one run — own searches, memory, digest and branded REPORT per customer; 10 weekly ≈ $8.40 the first week with AI
Works with	Schedules, webhooks, Sheets, Slack, Teams, Telegram, Make, n8n, Zapier, MCP

🎯 What this Actor does

1688 product sourcing without the spreadsheet — find suppliers in China in English: it runs the 1688.com products scraper on your account and keeps, and charges, only products worth contacting.

🏭 1688 supplier vetting — the trust score

A 0–100 score from 1688's own signals: service scores, years, factory inspection, manufacturer, badges, repeat buyers and orders — each reason citing the value.

💰 Unit cost at your quantity, landed cost and margin

200 pieces on a 2–199 / 200–999 / 1000+ ladder cost ¥5.90, not the ¥5.20 headline. Freight, duty and import VAT start at dated estimates, in USD, EUR or GBP. The same product from several suppliers is grouped, the cheapest at your quantity marked.

🈯 1688 in English — search, titles, badges, cities

Your English search goes to 1688 in Chinese (silicone phone case → 硅胶手机壳); titles, badges and places come back in English. Off-topic listings (felt bag in a felt-sheet search) are dropped before you pay.

✉️ A supplier message — and the shop to send it to

Every shortlist and maybe row gets an RFQ in Chinese and English next to the supplier's 旺旺 chat link. The Actor never contacts anyone.

🚀 How to use it

1. Click Try it — the form runs the free demo on 6 fictional sample listings as is.
2. Type your product into "Search 1688 for", your quantity and sell price — it replaces the sample and searches live on your Apify account, capped by "Max source spend". Leave "AI assessment" on.
3. Press Start — the Shortlist lands best first; Supplier messages has the link and the message to paste.
4. Schedule it weekly with your address in "E-mail me the new products" — only new products are charged.

📚 Learn more: Step-by-step tutorial — how to find 1688 suppliers with this Actor

🆚 Why this instead of running devcake's 1688 scraper?

	1688.com Products Scraper (devcake/1688-com-products-scraper)	1688 Supplier Finder
Your time per 100 products	≈ 13 h — translate, read tiers, check suppliers	≈ 0 h — ranked, costed, translated
Same 100 qualified products, all in	$0.50 for 250 unsorted raw rows + those 13 h	100 ranked, explained products: $1.10 rules only ($0.60 here + the same scan) · $2.11 with AI; junk costs $0
What you pay for	every row, junk included · 319 users / 30 days	qualified products only, first 25 free
Verdict, cost, language	none; Chinese (its README)	supplier score, MOQ-tier landed cost, margin, RFQ

vs other 1688 supplier finders

What no rival does: a supplier score, the unit cost at your MOQ tier with landed cost and margin, an RFQ per product — and you pay only for products that pass. English titles alone are not unique (webdata_labs, crawleast).

Option	What you get	Price	100 clean products
This Actor	scored, MOQ-tier costed, RFQ, translated	$0.006 / $0.015 per qualified	$1.10 rules only · $2.11 with AI
zen-studio/1688-wholesale-scraper (1,748 users)	50+ raw fields	$0.00499 a row + $0.0049 a run	≈ $1.25 + vetting by hand
webdata_labs/1688-scraper (113 users)	raw rows with English titles, MOQ tiers	$0.005 a row	≈ $1.25, unscored, no costing
crawleast/1688-image-search-scraper (233 users)	photo search; English titles, landed cost optional	$0.005 + $0.003 detail	photo only, unscored
om_kh/1688-reverse-image-product-finder · devcake/scraper-by-image (150 users)	1688 matches of a photo	per result	photo matches, no costing
quanmatrix/1688-supplier-sourcing-intelligence	supplier signals	$0.006 a row	≈ $1.50, untranslated
AliPrice (browser extension)	image search while browsing	free	one at a time

💸 A Jungle Scout (Supplier Database) alternative for 1688: $49 a month per added seat (list price, checked 2026-10-15), Alibaba only; a weekly 3-search monitor here ≈ $2.50 the first week, all-in.

📦 Already scraping 1688? Vet your 1688 scraper data

Ran zen-studio's or devcake's 1688 scraper already? Set "Source" to dataset and paste that run's dataset ID — no second scrape: supplier score, MOQ-tier landed cost, English titles and the RFQ are added (rules only by default), $0.006 per product that qualifies. Chain it after every scrape: Integrations.

💵 Pricing — what a run really costs

Pay per result: one event per qualified product, nothing else. Your first 25 qualified products are free; up to 5 of them get a live AI verdict on us.

Event	When it is charged	Price
free-tier	your first 25 qualified products on this Actor, in any mode	$0.00
qualified-product-basic	AI off — supplier score, costs, glossary title, template message	$0.006
qualified-product-ai	AI on — plus AI English title, relevance check, sourcing note, risk flags, AI message	$0.015

Never charged by us: listings below the minimum score, off-topic, filtered out, excluded or received before; no start fee. The 1688 scan is not ours to waive: the scraper bills your account $0.002 per listing, capped by "Max source spend".

How that compares — $0.006 is the live Store median for "1688 supplier finder" (7 priced Actors, 2026-10-15), but per qualified product: one replaces ≈ 2.5 raw rows (≈ $0.0024 per raw-row equivalent) and ≈ 8 minutes of vetting (on 5 sample products).

Worked example — 1 search × 100 listings → 40 qualified: $0.20 scan + 40 × ($0.015 + ≈ $0.0011 tokens) = $0.84 all-in with AI, ≈ $0.021 per qualified product (with the $0.005 scan share); $0.44 with AI off. Your first run's first 25 are free: ≈ $0.47.

At volume — Sourcing agent, 10 clients in one weekly run: ≈ $8.40 all-in the first week, then only new products. FBA seller, 3 searches weekly: ≈ $2.50 the first week.

🤖 What the AI tier adds

The same sample product, AI on (qualified-product-ai, $0.015, list-mode run 2026-10-15) — a better title, a relevance check, a grounded note and the message, moving the verdict from maybe to shortlist:

{"titleEn": "Liquid silicone phone case for Apple 16/16Pro magnetic solid color supports custom LOGO", "titleSource": "ai", "keyAttributes": ["liquid silicone", "phone case", "for Apple 16/16Pro", "magnetic", "solid color", "supports custom LOGO"], "verdict": "shortlist", "ruleScore": 69, "fitScore": 74, "sourcingNote": "Manufacturer with 5 years on 1688 and composite score 4.3/5. MOQ of 50 pieces is within order quantity of 200 at ¥7.5/pc with 88.7% margin. Lacks 1688 factory inspection.", "supplierMessageEn": "Hello! I am interested in your liquid silicone phone case (https://detail.1688.com/offer/100000000003.html). We plan an initial order of 200 pieces at a target unit price of 7.5 CNY. Could you please provide a formal quote, a sample, and the lead time? Looking forward to your reply, thank you!", "aiModel": "google:gemini-3.6-flash"}

Each AI field is checked after the model answers: attributes must be characters of the Chinese title, the note and message quote only listing values, your quantity and target price, and the fit moves at most 15 points.

By hand that is ≈ 8 minutes per product on 5 sample products (2026-10-15). Without model access the demo shows labelled sample verdicts (aiModel: "sample"); your first real search gets up to 5 live AI verdicts on us.

⚙️ Input

Field	Default	What it does
queries	—	What to search on 1688, English or Chinese — typing it starts the live search
goalDescription	—	Your product brief, for the relevance check and the message
minScore	60	Decides what you pay: suppliers below it are not charged
orderQuantity	100	Picks the price tier
targetSellPriceUsd · sellCurrency	— · USD	Adds marginPct, in USD, EUR or GBP
freightUsdPerKg · dutyPct · importVatPct	7 · 15 · 0	Landed cost — dated estimates, labelled on each row
maxQualified	50	Hard cap on output and spend
maxDiscoveryChargeUsd	0.5	Stops the 1688 scraper at this cost — form and API alike
maxProductsPerQuery	100	Listings read per search ($0.20 on your account)
enableAi	true	AI assessment on / off

📤 Output

One dataset item per qualified product — JSON, CSV or Excel, or a webhook. Views: Shortlist, Costs & margin, Supplier check, Supplier messages. Sort by score; ruleScore is the supplier part, fitScore the AI's — trend across runs on ruleScore.

Upsert key: dedupeKey — offer:<1688 offer id>: one product is one row however many runs find it. Records: OUTPUT (= RUN_SUMMARY: the funnel, charges, sameProductGroups, skipped), a DIGEST and a client-ready REPORT page.

⭐ Found it useful? A review on the Store helps others find it — it takes a minute on the Actor's page.

🔁 Run it weekly

1. Open the saved task and click Schedule after your first real run (OUTPUT.scheduleTaskUrl), your address in notifyEmail: each run with new products e-mails the digest with the trend vs the last run.
2. Keep dedupeAcrossRuns: true — only new products are charged; adding a search term keeps the memory.
3. Or a Slack, Teams, Google Chat, Discord or Telegram URL in sendAlertsTo — start from the ready tracker sheet.

Weekly cost is new products only: 3 searches ≈ $2.50 the first week with AI; week 2 (not measured yet) delivers less. Price-change alerts on your shortlist: roadmap 2026-11-15; rolling discovery 2026-11-05; supplier replies (a pasted 旺旺 quote → price, MOQ, lead time in the row) 2026-11-12.

🏢 Agency in 3 steps

Pick Sourcing agency in "Start from a ready niche search" (2 example clients), replace them with yours (name, queries, optionally quantity, sell price, notifyEmail), start and schedule the saved task: each client gets only its new products, its own e-mail and a REPORT-<client> page with its name.

🎯 Try it for your niche

Pick it in Start from a ready niche search (preset) and press Start — its first run is capped at $0.15 (35 listings per search) until a live run has measured its yield.

Niche	What it looks for	Start
Amazon FBA phone accessories	magnetic cases + stands, 300 pcs, $19.99, factories	fba-phone-accessories (input)
Pet store	dog chew toys + cat scratchers, 200 pcs, $12.99	pet-supplies (input)
Fitness brand	yoga mats + resistance bands, 500 pcs, sea freight	yoga-fitness (input)
EU / UK Amazon seller	baking mats + spatulas, €14.99, 19 % VAT	eu-amazon-seller (input)
Sourcing agency	2 example clients, each its own search and REPORT	agency-roster (input)

🔌 Integrations, automation and API

Every run writes OUTPUT.mode (demo, live, list, dataset) and OUTPUT.ok / outcome — branch on those.

• After every zen-studio/1688-wholesale-scraper or devcake/1688-com-products-scraper run: in its Integrations tab add "Run succeeded → start this Actor" with sourceMode: dataset, datasetId = the run's resource.defaultDatasetId, enableAi: false — every scrape is vetted, rules only until you switch AI on.
• Webhook — any https URL in sendAlertsTo gets each product as JSON. Sheets / Slack — the Integrations tab. Airtable / HubSpot — a Make / Zapier step, upsert on dedupeKey (native push: roadmap 2026-11-30).

Run outcomes

MCP (Claude, ChatGPT, any client) — https://mcp.apify.com/?actors=rich_minds/1688-supplier-finder-ai.

List-mode rows — what a pasted record needs

👥 Who is it for?

You are…	You run it to…	Start with
An Amazon FBA, TikTok Shop or Shopify seller	find factories that accept your quantity and leave your margin	queries, orderQuantity, targetSellPriceUsd
An EU / UK Amazon seller	the same, with landed cost in EUR / GBP and import VAT	the eu-amazon-seller preset
A sourcing agent or buying office	each Monday's new trusted suppliers per client	the agency-roster preset, weekly
A dropshipper	factory direct products with single-piece dropshipping	targetFlags: ["dropshipping"]
An AI agent or automation	a brief → a typed, costed shortlist with an RFQ	the minimal agent input

🧠 How the AI works

• Grounded: one listing at a time, only the rule shortlist (up to maxQualified).
• No keys required where your Apify plan includes model access — otherwise your own key (llmProvider: "byok"). Default google/gemini-2.5-flash-lite (a short JSON verdict, ≈ $0.0011 of tokens per product); with your key its vendor picks the model (Gemini: google:gemini-3.6-flash, as in the sample), its thinking capped.
• Fallback: a model failure keeps the rule score, title, note and message — billed as basic.

🔒 Data, compliance and limits

• Public 1688 offer and company data only — no login, no personal data; 1688's terms apply.
• Landed cost is an estimate (freight, duty, VAT, the CNY rate), not a freight quote.

❓ FAQ

How much will one run cost me? One search × 100 listings → ≈ 40 qualified: $0.44 rules only, $0.84 all-in with AI ($0.20 of it the scan). Your first 25 qualified products are free.

What will my first real search cost? The $0.20 scan, 15 products past the free 25 and tokens: ≈ $0.47 with AI.

Is this a 1688.com products scraper? It runs one on your account and adds vetting, costing, translation and the message — or vets 1688 scraper data you have (datasetId).

Can I search 1688 in English? Yes — it is sent to 1688 in Chinese (OUTPUT.queriesLocalized); Chinese finds most.

Does it work for EU / UK Amazon sellers? Yes: pick EUR or GBP in "Sell currency", add your import VAT and duty — landed cost and margin come in your currency (the eu-amazon-seller preset).

How do I verify chinese suppliers with it? The score reads 1688's verified signals (service scores, 验厂, years, repeat buyers), each reason citing the value. Order samples first: it ranks, it does not certify.

Which Apify plan do I need? Any plan runs the demo and your own rows; the live search needs a plan that runs Store Actors. AI per plan: At a glance → Keys / setup.

How long does a run take? The demo 0.5 s; a live search of 100 listings ≈ 2 minutes (estimate).

Will I be charged for the same product twice? No, with dedupeAcrossRuns on — adding a search term keeps the memory.

What does "qualified" mean? Every free filter passed, at least your minimum score, on-topic, a verdict you asked for; what the AI drops is named in OUTPUT.skipped.

🧩 More Actors from the same developer

Not quite your use case? The same pay-per-qualified-result model, closest to this Actor's buyers first:

• Amazon Bestsellers Scraper + AI — find the Amazon products worth sourcing, then their factory here.
• Facebook Ad Library Winning Ads Finder — see which products competitors advertise longest.
• Xiaohongshu KOL Finder — reach Chinese consumers on RedNote.

🆘 Support

Something missing or wrong? Open an issue on the Actor's page — buyers' requests ship first.

📝 Changelog

One line per build, newest first — what changed for you.

• 0.1.5 (2026-10-16) — AI vetting ≈ $0.021 per product all in (was $0.034); costs in EUR / GBP with import VAT; the same product from several suppliers grouped; vet a 1688 dataset you already scraped.
• 0.1.4 (2026-10-16) — agencies run every client in one go; demo rows marked as samples; a message on every shortlist and maybe row.
• 0.1.1–0.1.3 (2026-10-15) — first release: supplier score, landed cost, English titles and searches, a bilingual message.