Semantic vector search
Search unstructured content by semantic meaning using vector embeddings, supporting natural-language queries over text, images, and audio.
MongoDB Vector Search
Freemium
MongoDB Vector Search lets developers store and search vector embeddings alongside operational data in MongoDB Atlas. It supports semantic, hybrid, recommendation, anomaly-detection, and conversational AI applications.
什么是 MongoDB Vector Search？
MongoDB Vector Search is a semantic search capability for MongoDB Atlas. It stores vector embeddings alongside operational data so applications can retrieve content by meaning across unstructured data such as text, images, and audio.
The product is designed for AI applications that need context from live application data. It supports natural-language retrieval, hybrid queries that combine semantic and lexical search with structured database operations, and native reranking for improving result relevance before information is passed to an LLM.
By keeping operational and vector data in MongoDB, teams can avoid maintaining a separate vector database and synchronization workflow. Automated Embedding can generate and synchronize embeddings as indexed operational data changes, while distributed search architecture and Search Nodes support independent scaling of search workloads.
MongoDB Vector Search 能做什么？
Automated Embedding
Define an index on live operational data and have MongoDB generate and synchronize vector embeddings as the source data changes. Embedding models from Voyage AI can be selected for quality, cost, and latency requirements.
Hybrid query workflows
Combine vector queries with metadata filters, graph lookups, aggregation pipelines, geospatial search, and lexical search within a single database workflow.
Native reranking
Apply Voyage AI reranker models within the MongoDB query engine so the most relevant results can be selected before they are sent to an LLM.
Independent search scaling
Use MongoDB’s distributed architecture and Search Nodes to scale vector-search resources separately from the core database workload.
Atlas operational capabilities
Keep vector and operational data in Atlas, where the product page identifies built-in security and high availability as part of the enterprise-ready deployment.
使用场景
“Conversational AI and retrieval-augmented applications”
Retrieve semantically relevant context from application data for conversational experiences, then use reranking to narrow the results supplied to an LLM.
“Recommendation engines”
Use vector representations to find content, products, or other records with similar meaning or characteristics rather than relying only on exact terms.
“Anomaly detection”
Compare vector representations to identify unusual patterns in operational or unstructured data, using database filters and aggregation when additional context is required.
“Hybrid enterprise search”
Combine natural-language relevance with keywords, metadata constraints, geospatial conditions, or graph relationships when a search experience needs more than semantic similarity alone.
“Multimodal content retrieval”
Search collections containing unstructured text, images, or audio through vector representations while keeping those records with related operational data.
常见问题
What does MongoDB Vector Search store and search?
It stores vector embeddings alongside operational data in MongoDB Atlas and searches by semantic meaning. The source page specifically describes unstructured data such as text, images, and audio.
Do teams need a separate vector database?
MongoDB positions Vector Search as a way to keep operational and vector data in one place, avoiding a separate database and the synchronization work between systems.
How are embeddings generated?
With Automated Embedding, a team defines an index on live operational data. MongoDB then generates and synchronizes vector embeddings as the data changes, using a selected Voyage AI embedding model.
Can vector search be combined with regular database queries?
Yes. Vector queries can be combined with metadata filters, graph lookups, aggregation pipelines, geospatial search, and lexical search in a hybrid workflow.
Where is MongoDB Vector Search available?
The product is presented as a MongoDB Atlas capability. The pricing page lists Search & Vector Search on AWS, Google Cloud, and Microsoft Azure, and the home page also states that Search & Vector Search are available in Enterprise Advanced and Community Edition.
快速信息
- Product category
- Vector search and semantic retrieval
- Platform
- MongoDB Atlas
- Primary users
- Developers building AI, search, recommendation, and retrieval applications
- Supported search approach
- Semantic vector search, lexical search, and hybrid queries
- Embedding and reranking
- Voyage AI models through Automated Embedding and native reranking
- Cloud availability
- AWS, Google Cloud, and Microsoft Azure; pricing reference is region-specific
MongoDB Vector Search 替代品
Vespa.ai
vespa.ai
Vespa is an AI search platform for building search, retrieval-augmented generation, recommendation, personalization, and agent applications over text, vectors, tensors, and structured data. It is designed for developer teams that need configurable ranking and distributed operation at production scale.
turbopuffer
turbopuffer.com
turbopuffer is an object-storage-native search engine for vector, full-text, hybrid, and regex search. It helps AI, search, and data teams retrieve content from large document collections with filtering, ranking, and scalable namespace-based storage.
Supabase
supabase.com
Supabase is a Postgres development platform for building applications with a database, authentication, APIs, realtime features, serverless functions, storage, and vector support. It helps developers move from application code to a managed backend through a dashboard, client libraries, CLI, and generated APIs.
Upstash Vector
upstash.com
Upstash Vector is a serverless vector database for storing and querying embeddings and metadata in AI and machine-learning applications. It supports dense, sparse, and hybrid indexes through a REST API and SDKs for Python, TypeScript, Go, and PHP.
Voyage AI
www.voyageai.com
Voyage AI provides embedding models and rerankers for improving search and retrieval in AI applications. It is designed for teams building retrieval-augmented generation and other applications that use unstructured data.
Powabase
powabase.ai
面向 AI 应用的后端平台，集成 Postgres、身份验证、存储、实时功能、检索和智能体