# Pinecone — Managed Vector Database for Production AI

> Fully managed vector database for production AI search. Pinecone offers serverless scaling, hybrid search, metadata filtering, and enterprise security with zero infrastructure.

## Install

Copy the content below into your project:

## Quick Use

```bash
pip install pinecone
```

```python
from pinecone import Pinecone, ServerlessSpec

pc = Pinecone(api_key="...")

# Create serverless index
pc.create_index(
    name="docs",
    dimension=1536,
    metric="cosine",
    spec=ServerlessSpec(cloud="aws", region="us-east-1"),
)

index = pc.Index("docs")

# Upsert vectors
index.upsert(vectors=[
    {"id": "doc1", "values": [0.1, 0.2, ...], "metadata": {"title": "AI Guide"}},
    {"id": "doc2", "values": [0.3, 0.4, ...], "metadata": {"title": "ML Intro"}},
])

# Query
results = index.query(vector=[0.15, 0.25, ...], top_k=5, include_metadata=True)
for match in results["matches"]:
    print(f"{match['id']}: {match['metadata']['title']} ({match['score']:.4f})")
```

## What is Pinecone?

Pinecone is a fully managed vector database designed for production AI applications. Unlike self-hosted alternatives, Pinecone handles all infrastructure — scaling, replication, security, and updates. Its serverless architecture means you pay only for what you use, with automatic scaling from zero to billions of vectors.

**Answer-Ready**: Pinecone is a fully managed serverless vector database. Zero infrastructure, automatic scaling, hybrid search (dense+sparse), metadata filtering, and enterprise security. Used by thousands of companies for production RAG and search. Free tier with 100K vectors.

**Best for**: Teams wanting production vector search without managing infrastructure. **Works with**: OpenAI, Cohere, HuggingFace, LangChain, LlamaIndex. **Setup time**: Under 2 minutes.

## Core Features

### 1. Serverless (Zero Ops)

```python
# Create index — no clusters, no replicas
pc.create_index(
    name="my-index",
    dimension=1536,
    metric="cosine",
    spec=ServerlessSpec(cloud="aws", region="us-east-1"),
)
# Scales automatically. Pay per query + storage.
```

### 2. Metadata Filtering

```python
results = index.query(
    vector=[...],
    top_k=10,
    filter={
        "category": {"$eq": "technology"},
        "year": {"$gte": 2024},
        "tags": {"$in": ["ai", "ml"]},
    },
)
```

### 3. Namespaces (Multi-Tenancy)

```python
# Separate data by tenant
index.upsert(vectors=[...], namespace="tenant-a")
index.upsert(vectors=[...], namespace="tenant-b")

# Query within namespace
results = index.query(vector=[...], namespace="tenant-a", top_k=5)
```

### 4. Hybrid Search (Sparse + Dense)

```python
# Combine keyword and semantic search
results = index.query(
    vector=[...],         # Dense vector
    sparse_vector={"indices": [1, 5], "values": [0.5, 0.3]},  # Sparse
    top_k=10,
)
```

### 5. Integrated Inference

```python
# Pinecone generates embeddings for you
pc.inference.embed(
    model="multilingual-e5-large",
    inputs=["What is AI?"],
    parameters={"input_type": "query"},
)
```

## Pricing

| Tier | Vectors | Price |
|------|---------|-------|
| Free | 100K | $0 |
| Starter | 1M | From $8/mo |
| Standard | 10M+ | Usage-based |
| Enterprise | Unlimited | Custom |

## Pinecone vs Self-Hosted

| Aspect | Pinecone | Qdrant/Milvus |
|--------|----------|---------------|
| Setup | 2 minutes | Docker/K8s |
| Scaling | Automatic | Manual |
| Maintenance | Zero | You manage |
| Cost (small) | Free tier | Free (OSS) |
| Cost (large) | Higher | Lower (self-hosted) |
| SLA | 99.99% | Your responsibility |

## FAQ

**Q: When should I use Pinecone vs self-hosted?**
A: Pinecone for teams that want zero ops. Self-hosted (Qdrant, Milvus) for teams that want full control and lower costs at scale.

**Q: Does it support LangChain?**
A: Yes, first-class integration via `langchain-pinecone` package.

**Q: Can I migrate from Pinecone to self-hosted later?**
A: Yes, export vectors via the fetch API and import into any other vector database.

## Source & Thanks

> Created by [Pinecone](https://www.pinecone.io).
>
> [pinecone.io](https://www.pinecone.io) — Managed vector database

<!-- ZH -->

## 快速使用

```bash
pip install pinecone
```

两分钟启动全托管向量数据库。

## 什么是 Pinecone？

全托管 Serverless 向量数据库，零运维自动扩缩容，用于生产级 AI 搜索和 RAG。

**一句话总结**：全托管 Serverless 向量数据库，零运维 + 自动扩缩 + 混合搜索 + 元数据过滤 + 企业安全，免费层 100K 向量。

**适合人群**：不想管基础设施的 AI 搜索团队。

## 核心功能

### 1. Serverless — 零运维，按用量付费
### 2. 元数据过滤 — 复杂条件筛选
### 3. 命名空间 — 多租户隔离
### 4. 混合搜索 — Dense + Sparse

## 常见问题

**Q: Pinecone vs 自托管？**
A: Pinecone 零运维，自托管（Qdrant/Milvus）更便宜更灵活。

## 来源与致谢

> [pinecone.io](https://www.pinecone.io) — 全托管向量数据库

---
Source: https://tokrepo.com/en/workflows/0fc5f7e8-439d-414f-bdaf-b09e05e1af49
Author: AI Open Source