December 2025
Announcement Type | Description | Release date | Publishing Regions | References |
New feature | The Fields field now supports four new data types: ARRAY_STRING, ARRAY_INT, ARRAY_FLOAT, and ARRAY_LONG. | 2025-12-09 | Shanghai Shenzhen Beijing Hangzhou Zhangjiakou |
August 2025
Announcement Type | Description | Release date | Region | References |
New specifications | Five new instance types are added for storage-optimized clusters. | 2025-08-21 | Shanghai Shenzhen Beijing Hangzhou Zhangjiakou |
March 2025
Announcement Type | Description | Release date | Available regions | References |
New feature | The Fields field now supports the Long data type. | 2025-03-17 | Shanghai Shenzhen Beijing Hangzhou Zhangjiakou |
October 2024
Announcement Type | Description | Release date | Region | References |
New feature | Added the multi-vector retrieval feature. | 2024-10-09 | Shanghai Shenzhen Beijing Hangzhou Zhangjiakou |
July 2024
Announcement Type | Description | Release date | Region | References |
New regions available | The Serverless specifications for DashVector are now available in the Shanghai, Shenzhen, and Beijing regions. | 2024-07-10 | Shanghai Shenzhen Beijing | |
New specifications | DashVector now offers Serverless specifications. | 2024-07-08 | Hangzhou Zhangjiakou |
April 2024
Announcement Type | Description | Release date | Publishing Region | References |
New regions available | The service is now available in the Shanghai and Shenzhen regions. | 2024-04-26 | Shanghai Shenzhen | |
New region available | The service is now available in the Zhangjiakou region. | 2024-04-25 | Zhangjiakou | |
New feature | Added the feature to access the service from a virtual private cloud (VPC) using an internal same-region endpoint. | 2024-04-22 | Hangzhou Beijing |
March 2024
Announcement Type | Description | Release date | Region | References |
New region available | The service is now available in the Beijing region. | 2024-03-29 | Beijing | |
New feature | Added the group query feature. | 2024-03-25 | Hangzhou |
|
New feature | Added the access control feature for cluster instances. | 2024-03-25 | Hangzhou | |
Feature optimization | Optimized document operations:
Note Requirement: Python SDK version 1.0.13 or later. Java SDK version 1.0.11 or later. | 2024-03-25 | Hangzhou |
January 2024
Announcement Type | Description | Release date | Region | References |
New tutorials | Added tutorials on how to generate text embeddings, image embeddings, and multimodal embeddings. You can generate embeddings by calling an API or using an open source model. | 2024-01-17 | Hangzhou | |
New feature | Added the dynamic vector quantization feature. | 2024-01-23 | Hangzhou |
December 2023
Announcement Type | Description | Release date | Region | References |
Commercial release | Vector Retrieval Service DashVector is now commercially available. The service is sold based on cluster instances. You can create a free trial cluster to try out and test the product. | 2023-12-15 | Hangzhou |
October 2023
Announcement Type | Description | Release date | Region | References |
New feature | The Java SDK is officially released. | 2023-10-19 | Hangzhou | |
New feature | Added the feature to perform match queries using filter conditions. | 2023-10-19 | Hangzhou | |
New feature | Added support for sparse vectors. You can use sparse vectors to perform keyword-aware vector retrieval. | 2023-10-20 | Hangzhou |
September 2023
Announcement Type | Description | Release date | Region | References |
Public preview release | Vector Retrieval Service DashVector is now in public preview. The service is free of charge during the public preview period. Access through unencrypted connections is permanently disabled. | 2023-09-18 | Hangzhou | |
New feature | Added the feature to perform similarity searches based on a primary key (and its corresponding vector). | 2023-09-26 | Hangzhou |
August 2023
Announcement Type | Description | Release date | Region | References |
Invitational preview release | Vector Retrieval Service DashVector is now in invitational preview. DashVector is built on the Proxima kernel, an efficient vector engine developed in-house by Tongyi Lab. It provides a cloud-native, fully managed vector retrieval service with horizontal scalability. | 2023-08-11 | Hangzhou |