# Managed Vector Databases & AI Search Infrastructure — Comparison Matrix

> Dense, machine-readable comparison of 5 managed vector database platforms:
> vector limits, indexing, security/compliance posture, pricing and operational limits.
> Column headers link to the vendor via the `/go/<vendor>` affiliate redirect.
Last verified: 2026-10-09

## 1. Dimensions & Indexing

| Dimension | [Pinecone](/go/pinecone) | [Qdrant Cloud](/go/qdrant) | [Weaviate Cloud](/go/weaviate) | [Milvus (Zilliz Cloud)](/go/milvus) | [Supabase pgvector](/go/supabase) |
| :-- | :-- | :-- | :-- | :-- | :-- |
| **Max vector dimensions** | 20000 | 65535 | 65535 | 32768 | 16000 |
| **Index types** | serverless, pod-based | HNSW, FLAT (exact) | HNSW, flat (exact) | HNSW, FLAT, IVF_FLAT, IVF_SQ8, IVF_PQ, DiskANN, GPU indexes | HNSW, IVFFlat, exact scan |
| **Metadata filtering stage** | hybrid | pre-filtering | post-filtering | post-filtering | pre-filtering |
| **Quantization** | product, scalar, binary, int8 | scalar, product, binary | product, scalar, binary | scalar, product, binary | halfvec (binary/half precision via pgvector) |
| **Hybrid (dense + sparse) search** | ✅ yes | ✅ yes | ✅ yes | ✅ yes | ✅ yes |
| **Open-source core** | ❌ no | ✅ yes | ✅ yes | ✅ yes | ✅ yes |

## 2. Security & Compliance

| Requirement | [Pinecone](/go/pinecone) | [Qdrant Cloud](/go/qdrant) | [Weaviate Cloud](/go/weaviate) | [Milvus (Zilliz Cloud)](/go/milvus) | [Supabase pgvector](/go/supabase) |
| :-- | :-- | :-- | :-- | :-- | :-- |
| **SOC 2 Type II** | ✅ yes | ✅ yes | ✅ yes | ✅ yes | ✅ yes |
| **HIPAA (BAA)** | ✅ yes | ❌ no | ❌ no | ✅ yes | ✅ yes |
| **ISO 27001** | ✅ yes | ❌ no | ❌ no | ✅ yes | ✅ yes |
| **Data-residency regions (count)** | 10 | 8 | 5 | 10 | 8 |
| **Clouds offered** | AWS, GCP, Azure | AWS, GCP, Azure | AWS, GCP | AWS, GCP, Azure | AWS |

<details>
<summary>Full region lists</summary>

- **Pinecone** (10): aws-us-east-1, aws-us-east-2, aws-us-west-2, aws-eu-west-1, aws-eu-central-1, aws-ap-southeast-1, aws-ap-northeast-1, aws-ap-south-1, azure-eastus2, gcp-us-central1
- **Qdrant Cloud** (8): aws-us-east-1, aws-us-west-2, aws-eu-west-1, aws-eu-central-1, aws-ap-southeast-1, aws-ap-northeast-1, gcp-us-central1, azure-west-europe
- **Weaviate Cloud** (5): aws-us-east-1, aws-us-west-2, aws-eu-west-1, aws-ap-southeast-1, gcp-us-central1
- **Milvus (Zilliz Cloud)** (10): aws-us-east-1, aws-us-east-2, aws-us-west-2, aws-eu-west-1, aws-eu-central-1, aws-ap-southeast-1, aws-ap-northeast-1, aws-ap-south-1, azure-eastus, gcp-us-central1
- **Supabase pgvector** (8): aws-us-east-1, aws-us-east-2, aws-us-west-1, aws-us-west-2, aws-eu-west-1, aws-eu-central-1, aws-ap-southeast-1, aws-ap-northeast-1

</details>

## 3. Pricing & Commercial Terms

| Term | [Pinecone](/go/pinecone) | [Qdrant Cloud](/go/qdrant) | [Weaviate Cloud](/go/weaviate) | [Milvus (Zilliz Cloud)](/go/milvus) | [Supabase pgvector](/go/supabase) |
| :-- | :-- | :-- | :-- | :-- | :-- |
| **Free tier available** | ✅ yes | ✅ yes | ✅ yes | ✅ yes | ✅ yes |
| **Free tier quota** | Starter: 1 project, up to 200K vectors, 1 serverless index | Free forever: 1 project, 1 GB storage, community support | 14-day free trial cluster (1 GB) on Weaviate Cloud | Free Serverless: shared resources, limited request units, 1 project | Free plan: 2 projects, 500 MB Postgres with pgvector, shared compute |
| **Minimum monthly spend (paid)** | $0 (pay-as-you-go) | $0 (pay-as-you-go) | $0 (pay-as-you-go) | $0 (pay-as-you-go) | $25/mo |
| **Hourly / unit rate** | Serverless reads from $0.096/100K read units; pod-based from ~$0.129/hr (p1.x1) | Dedicated nodes from ~$0.067/hr; serverless usage-based (requests + storage) | Serverless: storage + operation based; dedicated pods billed hourly | Serverless: request units + storage; provisioned clusters billed hourly | Pro $25/mo includes compute; add-on compute billed hourly from ~$0.04/hr |

## 4. Operational Limits

| Limit | [Pinecone](/go/pinecone) | [Qdrant Cloud](/go/qdrant) | [Weaviate Cloud](/go/weaviate) | [Milvus (Zilliz Cloud)](/go/milvus) | [Supabase pgvector](/go/supabase) |
| :-- | :-- | :-- | :-- | :-- | :-- |
| **Request payload / batch** | Hard ceiling ~100 MB per upsert request; recommend <=2 MB query payloads | Large JSON payloads supported; chunk uploads above ~32 MB per point | Batch endpoint auto-chunks; no fixed per-request ceiling documented | Max vector 32,768 dims; ~64 MB per gRPC/REST message (server default) | Standard Postgres/PostgREST limits; batch INSERTs sized by statement memory |
| **API / query timeout** | HTTP 429 on overload; API requests time out around 60 s | Per-request timeout parameter; long searches capped by client deadline | Server-side query timeout configurable; client deadline recommended | No fixed server timeout; 60 s client deadline recommended for searches | PostgREST bounded by statement_timeout; use async jobs for heavy work |
| **Rate limiting** | Per-second request ceilings by scale/pod size; 429 + Retry-After on bursts | Plan-based requests/second with burst headroom; 429 + Retry-After | Plan-based request quotas; 429 + Retry-After on bursts | Request-unit metering on serverless; 429 + Retry-After on quota exhaustion | Per-plan API throttling (Free tier limits anon requests per IP) |
| **Spec last verified** | 2026-10-08 | 2026-10-08 | 2026-10-08 | 2026-10-08 | 2026-10-08 |

---

*Generated from [`data/vendors.json`](/content/index.md) by `scripts/generate_markdown.py`.
Specs are extracted weekly from vendor pricing/docs pages by `scripts/extract_specs.py`
(Groq `openai/gpt-oss-120b` structured outputs). Values are planning guidance — confirm
contractual figures against the vendor pricing pages before procurement.*