Pinecone All-in-One Integration

The only Pinecone actor you'll ever need. Crawl websites, store datasets, run semantic searches, delete vectors, and check index stats — all from a single tool. No OpenAI key required.

---

What You Get

Connect your Pinecone vector database to the web in minutes. Give us a URL (or a sitemap, or an existing dataset) and we handle everything: crawling, content extraction, chunking, embedding, and storage. Then search your data with natural language — right from the same actor.

5 modes. 1 actor. Zero unnecessary API keys.

---

Features

🔍 Crawl & Store

Point it at any website. The actor crawls pages with a real browser (Playwright), converts HTML to clean Markdown, splits content into smart chunks, and stores everything as vectors in your Pinecone index. Supports multiple start URLs and XML sitemaps.

📦 Store Dataset

Already have data from another Apify scraper? Skip the crawl — load any dataset directly into Pinecone. Works with any actor output.

🔎 Semantic Search

Ask questions in plain English. Get ranked results with scores, metadata, and full chunk text. Add metadata filters and optional reranking (BGE Reranker v2 M3 or Pinecone Rerank v0) for precision results.

🗑️ Delete Vectors

Remove specific vectors by ID, wipe everything matching a metadata filter, or clear an entire namespace. Full control over your index.

📊 Index Stats

Check dimensions, record count, metric type, and per-namespace breakdown instantly. No Pinecone Console needed.

---

What You Need

Your Pinecone API Key (required)

This actor stores and retrieves data from your Pinecone index — you bring your own key.

1. Sign up for free at app.pinecone.io — no credit card needed
2. Go to API Keys in the left sidebar
3. Copy your key and paste it into the pineconeApiKey field

💡 Pinecone's free plan includes 1 index and ~100,000 vectors — more than enough to get started.

OpenAI or Cohere Key (optional)

Only needed if you choose OpenAI or Cohere as your embedding provider. With Pinecone Integrated Inference (the default), no extra API key is required at all.

---

Quick Start

Crawl a website and store in Pinecone

{
    "mode": "crawl-and-store",
    "pineconeApiKey": "YOUR_PINECONE_API_KEY",
    "pineconeIndexName": "my-docs",
    "startUrls": [{ "url": "https://docs.example.com" }],
    "maxPages": 50,
    "embeddingProvider": "pinecone"
}

That's it. No OpenAI key. The index is created automatically.

Search your data

{
    "mode": "query",
    "pineconeApiKey": "YOUR_PINECONE_API_KEY",
    "pineconeIndexName": "my-docs",
    "queryText": "How do I configure authentication?",
    "topK": 5
}

Store an existing Apify dataset

{
    "mode": "store-dataset",
    "pineconeApiKey": "YOUR_PINECONE_API_KEY",
    "pineconeIndexName": "my-docs",
    "datasetId": "YOUR_DATASET_ID",
    "datasetFields": ["text"],
    "metadataFields": { "pageUrl": "url", "pageTitle": "title" }
}

Delete old vectors by URL

{
    "mode": "delete",
    "pineconeApiKey": "YOUR_PINECONE_API_KEY",
    "pineconeIndexName": "my-docs",
    "deleteMode": "by-filter",
    "deleteFilter": { "url": { "$eq": "https://docs.example.com/old-page" } }
}

---

Embedding Providers

Provider	Extra Key Needed	Default Model	Cost
Pinecone (recommended)	❌ No	multilingual-e5-large	Included in Pinecone plan
OpenAI	✅ Yes	text-embedding-3-small	~$0.02 / 1M tokens
Cohere	✅ Yes	embed-english-v3.0	~$0.10 / 1M tokens

---

Pricing

Event	Price
Page crawled & stored in Pinecone	$0.01
Dataset item stored in Pinecone	$0.005
Semantic search query	$0.02

Delete and Index Stats modes are free — no charge for management operations.

💡 Example: Crawl 200 pages = $2.00 · Run 50 searches = $1.00

---

How We Compare

Feature	Official Apify Integration	Other Actors	This Actor
Crawl + Store + Query + Delete + Stats	❌	❌	✅ 5 modes
No OpenAI key needed	❌	❌	✅
Semantic search built-in	❌	❌	✅
Delete by filter or ID	❌	❌	✅
Auto-create index	❌	Partial	✅
Dimension validation	❌	❌	✅
Reranking support	❌	❌	✅
Sitemap support	❌	❌	✅

---

Output

Crawl & Store / Store Dataset

Every chunk pushed to the dataset:

Field	Description
vectorId	Deterministic SHA-256 ID
url	Source page URL
title	Page title
chunkIndex	Chunk number
totalChunks	Total chunks from this page
chunkText	Full text content
contentHash	SHA-256 for change detection
storedAt	Timestamp

Query

Field	Description
id	Vector ID
score	Relevance score (0–1)
metadata	Stored metadata + chunk text

---

Tips

• Same URL = same vector ID. Re-crawling overwrites the same vectors — safe to re-run anytime.
• Large sites? Use sitemapUrl to crawl complete websites from one XML sitemap.
• Dynamic content? renderJavaScript: true (default) handles React, Next.js, and SPAs.
• Chaining scrapers? Run any scraper first, then use store-dataset mode with its dataset ID.
• Cost-free embeddings? Keep embeddingProvider: "pinecone" — no extra API costs.