Product characteristics

Embedding function

Data writing/retrieval is automatically vectorized to align with the use experience of traditional databases. Users do not need to pay attention to the vector generation process, which greatly reduces the threshold of use.

End to end AI kit

The end-to-end solution for content retrieval in the RAG field integrates the functions of document pre-processing, automatic vectorization, retrieval and fine sorting in the vector database, simplifies the data processing and retrieval process in RAG, and improves the efficiency of data access.

High performance

The vector database Tencent Cloud VectorDB single index supports a vector data scale of 100 billion levels, and can support millions of QPS and millisecond level query latency.

High availability

The vector database Tencent Cloud VectorDB provides multi copy high availability features, improves disaster tolerance, and ensures that the database can still operate normally in the face of node failures, load changes and other challenges.

low cost

Simply follow the instructions of the console to quickly create vector database instances. The whole process platform is managed without any installation, deployment, operation and maintenance operations, reducing machine costs, operation and maintenance costs, and labor costs.

Stable and reliable

The vector database Tencent Cloud VectorDB is derived from the vector retrieval engine OLAMA developed by Tencent Group itself. Nearly 40 business lines are running stably, and the daily average number of search requests is up to 100 billion. The service continuity and stability are guaranteed.

Product specification

Free Admission

Free beta

Guangzhou

  • CPU

    1 core

  • Memory

    1G

  • disk

    20GB

  • Number of nodes

    1
  • Trial scenario: only for quick test
High cost performance

Stand alone version (computational)

Guangzhou

  • CPU

    2 cores

  • Memory

    4G

  • disk

    20G

  • Number of nodes

    1
  • Recommended for individuals and small enterprises
  • Computational for QPS high and delay sensitive scenarios
  • Vector scale: 50w (1536 dimensions)
three hundred and twenty-four RMB/month
Buy Now
High cost performance

Stand alone version (storage type)

Guangzhou

  • CPU

    1 core

  • Memory

    8G

  • disk

    20GB

  • Number of nodes

    2
  • Recommended for individuals and small enterprises
  • Storage type is used in scenarios with large data volume and low QPS
  • Vector scale: 100w (1536 dimensions)
five hundred and eighty-four RMB/month
Buy Now
High cost performance

High availability version (storage type)

Guangzhou

  • CPU

    4-core

  • Memory

    32G

  • disk

    100GB

  • Number of nodes

    3
  • Recommended for medium and large enterprises
  • Support configuration of 3 data replicas
  • Vector scale: 400w (1536 dimensions)
three thousand five hundred and sixty-four RMB/month
Buy Now

End to end AI kit

AI suite is a one-stop document retrieval solution provided by Tencent Cloud VectorDB, which includes automated document parsing, information supplement, vectorization, content retrieval and other capabilities, and has a wealth of configurable items, which can significantly improve the recall effect of document retrieval. Users can quickly build a high-quality proprietary knowledge base within a few minutes by uploading the original documents, greatly improving the efficiency of knowledge access.

Embedding function

Better experience

Automatic vectorization of data writing/retrieval, aligned with the use experience of traditional databases, without paying attention to the vector generation process

Performance improvement

Optimize GPU processing speed, improve performance by 5~10 times, and improve efficiency

Shared resource pool

Massive GPU resource pool provides computing services, which can be used on demand with better cost

Product architecture

Tencent Cloud VectorDB is based on the vector engine OLAMA of Tencent Group, which processes hundreds of billions of searches every day. The bottom layer uses Raft distributed storage, and cluster management and scheduling are carried out through the master node to achieve efficient operation of the system. At the same time, Tencent Cloud Vector Database supports the setting of multiple shards and replicas, which further improves the load balancing capability, enabling the vector database to achieve high performance, high scalability and high disaster tolerance while processing massive vector data.

Application scenarios

  • Large model knowledge base
  • Recommendation system
  • Text/image retrieval

Tencent cloud vector database can be used together with the big language model LLM. After text segmentation and vectorization, the private domain data of enterprises can be stored in Tencent Cloud Vector Database to build an enterprise specific external knowledge base, so as to provide prompt information for the big model in subsequent retrieval tasks and assist the big model to generate more accurate answers.

According to our Getting Started Guide , you can create your first Tencent cloud vector database instance with just a few mouse clicks.
Go now