Neo4j is a graph database for knowledge graphs, fraud analysis and other connected data, run as the managed AuraDB service or self-hosted as open source.
Neo4j stores records as nodes linked by relationships, each carrying its own properties, so questions about how things connect are answered by following those links. Labels and properties can be added as the data grows, without defining a schema up front. Neo4j, Inc. develops and sells it. It also serves as a knowledge layer for AI applications that query the graph for connected facts. Queries are written in Cypher, a declarative language that matches patterns of nodes and relationships and aggregates results inside the database. Each change runs in an ACID transaction that commits or rolls back as a whole. Full-text indexes rank matches with Lucene relevance scores, and vector indexes search embeddings with a filter in the same query. AuraDB runs the database as a fully managed service with change data capture, read-only secondaries and point-in-time restore. Aura Graph Analytics opens analytics sessions over AuraDB, relational databases or CSV files with no ETL step. The database is built for developers and data teams whose questions depend on relationships rather than single rows. Typical owners are fraud and risk compliance teams, engineers grounding generative AI in a knowledge graph, and analysts running graph analytics. Security teams can restrict access by role, by property or by attribute, and sign users in through single sign-on. Teams that want no servers to manage pick AuraDB on AWS, Azure or Google Cloud. Teams that keep databases on their own hardware install the open source Community Edition, the Enterprise Edition or Infinigraph for sharded deployments. Common jobs include feeding a retrieval pipeline with facts from a knowledge graph and fraud and risk compliance work. Data arrives from Snowflake, BigQuery, Kafka or Spark through integrations, and change data capture passes node and relationship updates on to downstream consumers. Applications connect through drivers for Python, Java, JavaScript, Go and .NET, and AI tools can reach the graph through MCP. Engineers query it wherever the answer to a question is a path through connected records.
Pricing
View plans-
AuraDB Free
Free
- No credit card or other payment method required
- Access to all graph tools
-
AuraDB Professional
Custom pricing
- $0.09 per GB of instance memory per hour, minimum 1GB instance
- About $65.70 a month for a 1GB instance over 730 hours, computed from the hourly rate
- Up to 128GB memory per instance
- Storage, IO, backup and data transfer included
- Daily backups with 7-day retention
- Runs on AWS, Azure or Google Cloud
- Advanced instance-level metrics
- Trial available
- Sales tax may apply in MA, NY, TN, TX and WA
-
AuraDB Business Critical
Custom pricing
- $0.20 per GB of instance memory per hour, minimum 2GB instance
- About $292 a month for the 2GB minimum over 730 hours, computed from the hourly rate
- Highly available across multiple zones with a 99.95% uptime SLA
- Up to 2TB memory per instance on Google Cloud, 512GB on AWS and Azure
- Multiple databases per instance
- RBAC, PBAC and ABAC, instance-level SSO and IP filtering
- Daily backups with 30-day retention and hourly point-in-time restore
- Read-only secondaries and change data capture
- 24x7 support
- Trial available
-
AuraDB Virtual Dedicated Cloud
Custom pricing
- Price on request from Neo4j sales
- Everything in Business Critical in a dedicated environment
- VPC isolation and dedicated infrastructure
- Hourly backups with 60-day retention
- Enhanced 24x7 support
- Up to 2TB memory per instance on Google Cloud, 512GB on AWS and Azure
Features
- Included: Schemaless document storage
- Not included: Secondary indexes on any field
- Included: Multi-document ACID transactions
- Included: Automatic sharding across nodes
- Not included: Tunable consistency per operation
- Not included: Multi-region active-active writes
- Included: Server-side aggregation pipelines
- Included: Change streams for downstream consumers
- Included: Time-to-live expiry on records
- Included: Integrated full-text search on documents
- Included: Vector search on embedded fields
- Not included: Multi-model data access
- Included: Collection-level access roles
- Not included: Client-side field-level encryption
- Included: Continuous backup with point-in-time restore
- Not included: Embedded mobile database with sync
- Included: Open source community edition
- Not included: Serverless on-demand capacity pricing
Additional Features
- Cypher query language
- Property graph model with optional schema
- ACID transactions
- Full-text indexes with Lucene relevance scores
- Vector indexes with filtered search
- Change data capture on nodes and relationships
- Role, property and attribute-based access control
- Instance-level SSO and IP filtering
Best for
- Developers modelling data where relationships drive the queries
- Data teams building knowledge graphs for generative AI
- Fraud and risk compliance teams
- Analysts running graph analytics over warehouse, relational or CSV data
- Teams wanting a managed graph database on AWS, Azure or Google Cloud
- Teams that self-host and want an open source edition
- Enterprises that need attribute-based access control and single sign-on
Use cases
- Building a knowledge graph for generative AI retrieval
- Fraud and risk compliance analysis on connected records
- Graph analytics over AuraDB, relational or CSV data
- Vector similarity search inside graph queries
- Full-text search over node properties
- Streaming database changes to downstream systems
- Prototyping a graph model on a no-cost AuraDB instance
- Self-hosting a graph database under an open source license
- Restoring a database to an earlier point in time
- Expiring nodes automatically at a set timestamp
Screenshots & Videos
Explore Neo4j in action
User Reviews
Compare with Similar Software
Click to compare side by side with Neo4j
