Tencent Cloud VectorDB
What Tencent Cloud VectorDB is
Tencent Cloud VectorDB is a fully managed distributed database for storing, indexing, retrieving, and managing multidimensional embedding vectors. Its AI suite combines document parsing, enrichment, vectorization, and retrieval so users can create knowledge bases from uploaded documents. Text can be converted into vectors during insertion, updates, and similarity searches, with embedding charged by token volume. Search supports dense and sparse vectors, metadata filtering, and hybrid result merging through RRF or weighted ranking. Standard indexes include FLAT, HNSW, and IVF families; capacity-enhanced editions add DISK_FLAT from version 3.0. Tencent lists LLM knowledge bases, recommendation systems, and text or image retrieval as use cases, with RAG examples using DeepSeek, Hunyuan, and Baichuan. The cloud service offers API, Linux, and web access plus Python, Java, and Go SDKs. Tencent states one index supports up to hundreds of billions of vectors, million-level QPS, and millisecond query latency. Paid pricing starts at 344 CNY/mo, and a one-month free test instance is available.
Who it suits
VectorDB suits teams building LLM knowledge bases, recommendation systems, or text and image retrieval services. Its SDKs and cloud access support Python, Java, Go, API, Linux, and web workflows.
What it does well
- Managed distributed vector database
- Supports dense, sparse, and hybrid search
- Offers metadata filtering and multiple index families
- Provides Python, Java, and Go SDKs
- Includes one-month free test instance
What to know first
- Embedding usage is billed by token volume
- Free test instance lasts one month
- AI Suite similarity search accepts one text input
- Paid pricing starts at 344 CNY/mo
Limits
The free test instance runs on one node in one availability zone for one month, after which another instance must be purchased. AI Suite similarity search accepts one text input at a time, while regular vector or Embedding searches accept up to 20 inputs.
Plans
| Plan | Billed and limits | Price |
|---|---|---|
| 高可用版（计算型） | 1256元/月 · 广州 · 4核 CPU · 8G 内存 · 20GB 磁盘 · 2节点 · 200万向量（768维） | $1,256 a month |
| 高可用版（存储型） | 584元/月 · 广州 · 1核 CPU · 8G 内存 · 20G 磁盘 · 2节点 · 200万向量（768维） | $584 a month |
| 高可用版（标准型） | 872元/月 · 广州 · 2核 CPU · 8G 内存 · 20GB 磁盘 · 2节点 · 200万向量（768维） | $872 a month |
As cloud.tencent.cn lists them, read 9 Oct 2026. The maker's own page has the final word.
Getting started
- Visit https://cloud.tencent.cn/product/vdb
- Use the API or web access, or connect from Linux
- Select the one-month free test instance or a paid plan
- Connect with the Python, Java, or Go SDK
- Upload source documents through the AI suite to build a knowledge base
Verdict
Tencent Cloud VectorDB combines vector storage, retrieval, indexing, and an integrated RAG suite with several SDK options. Its scaling claims, security controls, and free test instance should be weighed alongside usage charges and plan pricing.
Questions
What does Tencent Cloud VectorDB cost?
Paid pricing starts at 344 CNY/mo. Listed plans include 584.00 USD, 872.00 USD, and $1,256 a month, each billed at the corresponding yuan amount shown.
Is there a free trial?
Yes. A free test instance runs on one node in one availability zone for one month.
Which platforms and SDKs are supported?
The service supports API, Linux, and web access. SDKs are available for Python, Java, and Go.
Which index types are available?
Index options include FLAT, HNSW, and the IVF family. The capacity-enhanced edition also supports DISK_FLAT from version 3.0.
Does it support hybrid or sparse search?
Yes. It supports sparse vectors and hybrid search combining dense and sparse retrieval, with RRF or weighted result merging.
What security controls are documented?
Documentation describes account and API-key authentication, VPC isolation, security-group traffic controls, and CAM-based resource permissions.
Facts from Tencent Cloud VectorDB's own pages
- Availability
- Tencent states an availability SLA of at least 99.97% for three-availability-zone high-availability deployments and 99.95% for two-zone deployments. · cloud.tencent.cn, 9 Oct 2026
- Embedding
- Embedding can automatically convert input text into vectors during insertion, updates, and similarity searches; usage is billed by token volume. · cloud.tencent.cn, 9 Oct 2026
- Free test instance
- Tencent describes a free test instance deployed on one node in one availability zone, limited to one month; continued use requires purchasing another instance. · cloud.tencent.cn, 9 Oct 2026
- Hybrid search
- Hybrid search combines dense and sparse vector retrieval and supports RRF and weighted result merging. · cloud.tencent.com, 9 Oct 2026
- Indexes
- The standard edition supports FLAT, HNSW, and IVF index families, while the capacity-enhanced edition supports DISK_FLAT disk indexes from version 3.0. · cloud.tencent.cn, 9 Oct 2026
- Limits
- The AI suite documentation says AI Suite similarity search accepts only one text input at a time, while regular vector or Embedding searches can accept up to 20 inputs. · cloud.tencent.cn, 9 Oct 2026
- Model integrations
- Tencent documents RAG examples using VectorDB with DeepSeek, Hunyuan, and Baichuan models. · cloud.tencent.cn, 9 Oct 2026
- Product
- Tencent Cloud VectorDB is a fully managed, self-developed distributed database for storing, indexing, retrieving, and managing multidimensional embedding vectors. · cloud.tencent.cn, 9 Oct 2026
- RAG suite
- Its AI suite combines document parsing, information enrichment, vectorization, and retrieval so users can upload source documents to build a knowledge base. · cloud.tencent.cn, 9 Oct 2026
- Scale
- The product page states that one index supports up to hundreds of billions of vectors, with up to million-level QPS and millisecond query latency. · cloud.tencent.cn, 9 Oct 2026
- SDKs
- Tencent provides Python, Java, and Go SDKs, and its connection guide specifies minimum versions of Python 3.6, Java 8, and Go 1.15. · cloud.tencent.cn, 9 Oct 2026
- Security
- The product documentation describes account and API-key authentication, VPC isolation, security-group traffic controls, and CAM-based resource permissions. · cloud.tencent.cn, 9 Oct 2026
- Support
- Tencent directs product questions to online customer service or support tickets and also points users to its developer community. · cloud.tencent.cn, 9 Oct 2026
- Use cases
- Tencent lists LLM knowledge bases, recommendation systems, and text or image retrieval as application scenarios. · cloud.tencent.cn, 9 Oct 2026
Where Tencent Cloud VectorDB stands on Freedom251
- Vector Databases#23 of 27