OpenSearch is an intelligent search platform that helps you build high-performance, real-time search services at low cost.
Introduction
OpenSearch is a one-stop, intelligent search platform built on Alibaba's proprietary large-scale distributed search engine. It powers the core search businesses of Alibaba Group, including Taobao, Tmall, and Cainiao. Battle-tested through years of industry use and traffic surges such as the 11.11 Global Shopping Festival, OpenSearch provides high-performance, real-time, highly available, and stable search services in six editions: , LLM-powered AI Chat Edition, Industry Algorithm Edition, OpenSearch High-performance Search Edition, Vector Search Edition, and Retrieval Engine Edition, to meet the search needs of various industries. By delivering professional search technology as a service, OpenSearch simplifies search, lowers the barrier to entry, and reduces costs. You can implement search features at low cost and iterate quickly.
To build a search service with OpenSearch, follow these steps:
-
Create an OpenSearch instance.
-
Edit and configure the instance.
-
Synchronize or upload documents.
-
Submit search requests from your app or website, and then call the API to retrieve search results from OpenSearch.
Service architecture
The AI Search Open Platform provides high-performance, component-based services for RAG and search scenarios. Its flexible component assembly mechanisms help you quickly build RAG and search workflows.
LLM-powered AI Chat Edition
A SaaS platform. Developers interact with the system through the console or an API.
Built on large language models (LLMs) and industry-leading NLP models, this edition lets you import business data to build an enterprise-specific large model for Q&A and search.
The Q&A search feature generates answers from your business data, delivering secure, stable, high-quality, and reliable results.
Industry Algorithm Edition
SaaS platform: Developers can interact with the system through the console or an API.
Scenario-specific configuration: Create an application instance, configure a data source, define the index schema and search properties, and wait for the index to be rebuilt. Then, you can test the search feature using the SDK or the console.
Industry templates: The Industry Algorithm Edition offers algorithm packages for industries such as e-commerce, content communities, gaming, and education. Select an industry vertical to create a full suite of search features, including query semantic understanding, sorting, and search guidance, for faster integration and effective intelligent search.
OpenSearch High-performance Search Edition
SaaS platform: Developers can interact with the system through the console or an API.
Scenario-specific configuration: Create an application instance, configure a data source, define the index schema and search properties, and wait for the index to be rebuilt. Then, you can test the search feature using the SDK or the console.
Big data retrieval: Unlike the Industry Algorithm Edition, this edition removes complex industry-specific algorithms and focuses on query and write throughput. It supports general search features such as analyzers and sorting, and provides responses within seconds and real-time queries for large-dataset retrieval scenarios.
Vector Search Edition
PaaS platform: Developers manage and maintain Vector Search Edition instances through a visual control platform.
Scenario-specific configuration: Create an application instance, configure a cluster, configure a data source, and set up an index table. After the index is rebuilt, you can test the search feature using the SDK or the console.
Vector search: Supports a wide range of query parameters and hybrid search with inverted and vector indexes. Provides high reliability and performance for image search, audio and video search, recommendations, and Q&A.
Retrieval Engine Edition
PaaS platform: Developers manage and maintain Retrieval Engine Edition instances through a visual control platform.
Scenario-specific configuration: Create an application instance, configure a cluster, configure a data source, and set up an index table. After the index is rebuilt, you can test the search feature using the SDK or the console.
SQL queries: Supports SQL query syntax, reducing the learning curve for developers.
Low cost: Includes multiple built-in index compression policies to reduce business costs while meeting query performance needs.