Skip to content
SPECIFICTION
Get listed

NebulaGraph

nebula-graph.io

Free plan Open source WebWindowsmacOSLinuxAPISelf-hosted
Visit nebula-graph.io
Spec score 8.5 #7 of 39 in Graph Databases

What it is

NebulaGraph is a distributed graph database for storing and querying complex connected data. It is designed for scalability and millisecond responses to complex queries; the project says it can store and process graphs with trillions of edges and vertices. It supports nGQL and openCypher compatibility, while native ISO-GQL is supported in Enterprise v5.0. Native vector search combines graph structures with unstructured data. Deployment options include on-premises, public cloud, and hybrid environments, and NebulaGraph Cloud can deploy clusters on Azure, AWS, and Google Cloud Platform. Developers can use Java, Python, C++, and Go clients, plus Spark and Flink connectors. NebulaGraph Studio is a web-based tool for schema creation, data import, query writing, and graph exploration. Cloud features include Private Link, role-based graph-space permissions, Raft-based redundant backups across server rooms in the same city, and point-in-time snapshots. The open-source edition is free under Apache 2.0; a 14-day Cloud trial includes one instance with 2 vCPUs, 4 GB RAM, and 10 GB storage without payment information.

Spec sheet

What it does
NebulaGraph is a distributed graph database for highly connected data, targeting millisecond queries at very large scale.Read 15 Oct 2026 · nebula-graph.io
Deployment
The maker supports on-premises, public-cloud and hybrid deployment models.Read 15 Oct 2026 · nebula-graph.io
Integrations
Official ecosystem tools include Java, Python, C++, and Go clients plus Spark and Flink connectors.Read 30 Sep 2026 · download.nebula-graph.io
Security
NebulaGraph Cloud provides Private Link for private network access and role-based access control for graph-space permissions.Read 30 Sep 2026 · download.nebula-graph.io
Support
NebulaGraph Cloud lets customers create work orders from the dashboard for assistance from a technical support specialist.Read 30 Sep 2026 · download.nebula-graph.io
License
The core NebulaGraph repository is licensed under Apache 2.0, allowing modification, deployment and commercial SaaS use.Read 15 Oct 2026 · github.com
Headquarters
Cupertino, California, United StatesRead 28 Sep 2026 · nebula-graph.io
Founded
2018Read 28 Sep 2026 · nebula-graph.io
Additional clients
PHP, Node.js, .NET, JDBC and several ORM clients are available but carry no uptime guarantee in the documentation.Read 15 Oct 2026 · docs.nebula-graph.io
API and SDKs
Official clients are available for C++, Java, Python, Go and a native console; Java is the only client documented as thread-safe.Read 15 Oct 2026 · docs.nebula-graph.io
Availability
NebulaGraph Cloud uses Raft for redundant backups across multiple server rooms in the same city.Read 30 Sep 2026 · download.nebula-graph.io
Cloud providers
NebulaGraph Cloud can deploy clusters on Microsoft Azure, AWS, and Google Cloud Platform.Read 30 Sep 2026 · download.nebula-graph.io
Cloud security
Cloud access controls can assign permissions by user, role, graph, space or query.Read 15 Oct 2026 · nebula-graph.io
Cloud service
NebulaGraph Cloud is a fully managed service with no-code setup, built-in demos, visual exploration and pay-as-you-go billing.Read 15 Oct 2026 · nebula-graph.io
Cloud trial
The cloud trial provides one 2-vCPU, 4-GB RAM and 10-GB storage instance for 14 days without payment information.Read 15 Oct 2026 · nebula-graph.io
Company founding
Vesoft says it was founded in October 2018.Read 15 Oct 2026 · nebula-graph.io
Company headquarters
Vesoft says it is headquartered in Cupertino, California.Read 15 Oct 2026 · nebula-graph.io
Company legal name
Crunchbase lists the legal name as Vesoft Inc and Nebula Graph as an alternate name.Read 15 Oct 2026 · crunchbase.com
Company size
Crunchbase lists Vesoft in the 101–250 employee range.Read 15 Oct 2026 · crunchbase.com
Consistency
Storage replication and high availability use the Raft consensus protocol.Read 15 Oct 2026 · github.com
Data import
The database supports diverse data formats and integration with existing ingestion pipelines.Read 15 Oct 2026 · nebula-graph.io
Distributed architecture
The system separates Graph, Storage and Meta services so compute and storage can scale independently.Read 15 Oct 2026 · docs.nebula-graph.io
Documentation depth
The documentation covers architecture, deployment, nGQL, clients, ecosystem tools, operations, FAQs and release notes, and the 3.8.0 manual was revised September 24, 2026.Read 15 Oct 2026 · docs.nebula-graph.io
Flink integration
The official Flink connector supports NebulaGraph 2.0 and 3.0 and requires Java 8 or newer.Read 15 Oct 2026 · github.com
Free trial limits
The 14-day NebulaGraph Cloud trial includes one database instance with 2 vCPUs, 4 GB RAM, and 10 GB storage, and requires no payment information.Read 30 Sep 2026 · nebula-graph.io
Funding
Vesoft’s news page reports an $8 million Series pre-A round led by Redpoint China Ventures on June 15, 2020.Read 15 Oct 2026 · vesoft.com
Graph algorithms
NebulaGraph advertises more than 50 graph algorithms and separate analytics capabilities.Read 15 Oct 2026 · nebula-graph.io
Graph model
NebulaGraph stores entities as vertices and relationships as typed edges with properties.Read 15 Oct 2026 · docs.nebula-graph.io
Graph scale
NebulaGraph states that it can store and process graphs with trillions of edges and vertices.Read 30 Sep 2026 · download.nebula-graph.io
Limit: cluster operations
A G2 reviewer said cluster setup and management for horizontal scaling can be complicated.Read 15 Oct 2026 · g2.com
Limit: missing temporal versioning
The HydraDB review reports that NebulaGraph lacks native temporal versioning, leaving historical-state tracking to application logic and schema design.Read 15 Oct 2026 · hydradb.com
Limit: operational complexity
An independent HydraDB review says self-managed deployments require planning for topology, capacity, monitoring, balancing, backup and recovery across services.Read 15 Oct 2026 · hydradb.com
Limit: query compatibility
The HydraDB review says nGQL compatibility with openCypher is partial, so migrations from Cypher systems may require query translation.Read 15 Oct 2026 · hydradb.com
Notable news: cloud GA
NebulaGraph announced general availability of the Cloud Standard Tier with a 14-day trial on January 15, 2025.Read 15 Oct 2026 · nebula-graph.io
Notable news: Enterprise 5.3
The company history page records NebulaGraph Enterprise v5.3 GA in June 2026 with built-in algorithms, subgraph queries, dynamic replica control, multi-tenancy and Dry Run mode.Read 15 Oct 2026 · nebula-graph.io
Praise: integration
Another G2 reviewer praised integration with big-data and enterprise software and fast real-time analysis.Read 15 Oct 2026 · g2.com
Praise: learning and scale
A G2 reviewer described NebulaGraph as easy to learn, highly scalable and performant on large datasets.Read 15 Oct 2026 · g2.com
Query language
The open-source repository describes an OpenCypher-compatible language, while the current product pages advertise native ISO-GQL support.Read 15 Oct 2026 · github.com
Query languages
NebulaGraph supports nGQL, openCypher compatibility, and native ISO-GQL support in Enterprise v5.0.Read 30 Sep 2026 · download.nebula-graph.io
Rating on G2
5.0 out of 5 from 2 reviewsRead 15 Oct 2026 · g2.com
Recovery
NebulaGraph supports snapshots that can restore an instance to a point in time and automatically replace failed instances.Read 30 Sep 2026 · download.nebula-graph.io
Security controls
The edition comparison lists transmission encryption, user and role management, graph-space permissions, node-edge permissions, attribute permissions and audit logs.Read 15 Oct 2026 · nebula-graph.io
Self-hosting
Users can build from source, deploy binaries on one or multiple machines, or run the database through cloud offerings.Read 15 Oct 2026 · github.com
Spark and Flink
The download site lists Spark and Flink connectors for data-science workflows.Read 15 Oct 2026 · download.nebula-graph.io
Storage recommendation
The manual recommends local NVMe SSDs, discourages HDD and remote NAS or SAN storage, and says RAID is unnecessary because replication is built in.Read 15 Oct 2026 · docs.nebula-graph.io
System requirements
Compilation requires x86_64, 4 GB RAM and 10 GB SSD; Linux with kernel 4.15 or newer is recommended.Read 15 Oct 2026 · docs.nebula-graph.io
Use cases
The maker presents fraud detection, recommendations, knowledge graphs, supply-chain management, master-data management and generative AI as target applications.Read 15 Oct 2026 · nebula-graph.io
Vector search
The current database product includes native vector search for graph and unstructured-data workflows.Read 15 Oct 2026 · nebula-graph.io
Visualization
NebulaGraph provides Studio, Explorer and Dashboard tools for importing, exploring and operating graph data.Read 15 Oct 2026 · vesoft.com

Plans and prices

Open-source Edition Free subset of core features · community collaboration and support Read 15 Oct 2026 · nebula-graph.io
NebulaGraph Cloud Standard Priced on request pay-as-you-go per hour; node, storage and transfer charges single node up to 8 vCPUs, 64 GB RAM and 320 GB storage · no high availability Read 15 Oct 2026 · nebula-graph.io
NebulaGraph Cloud Professional Priced on request pay-as-you-go per hour; storage-node, query-node, disk and transfer charges up to 10 query nodes and 10 storage nodes · high availability · faster technical support Read 15 Oct 2026 · nebula-graph.io
Enterprise Edition Priced on request contact sales enhanced performance · enterprise-grade stability and security · premium support Read 15 Oct 2026 · nebula-graph.io

Every NebulaGraph plan and price →

Strong points and limits

Strong points

  • Open-source Edition is free under Apache 2.0.
  • Supports nGQL and openCypher compatibility.
  • Native vector search combines graph and unstructured data.
  • Cloud trial needs no payment information.

Limits

  • Trial is limited to one database instance.
  • Standard Cloud plan has no high availability.
  • ISO-GQL support is limited to Enterprise v5.0.

Who it suits

NebulaGraph suits teams modeling connected data for use cases such as recommendations, fraud detection, knowledge graphs, and network analysis. Its deployment options cover on-premises, cloud, and hybrid environments.

NebulaGraph in depth

Read the full write-up

NebulaGraph is a distributed graph database for teams handling complex relationships at very large scale. It suits developers who need self-hosted, public-cloud, or hybrid deployment, plus graph and vector search in one system. The main choice is whether the free edition’s limits are acceptable or cloud high availability and enterprise capabilities are required.

Overview

NebulaGraph targets millisecond querying of complex connected data and states that it can store and process graphs containing trillions of vertices and edges. That ambition fits recommendation engines, social networks, fraud detection, cybersecurity, network analysis, knowledge graphs, and supply-chain systems. The scale is a strong reason to shortlist it for graph-heavy workloads, but deployment and edition limits matter more for smaller teams than the headline capacity.

Querying is available through nGQL and openCypher compatibility. Native ISO-GQL arrives in Enterprise v5.0, so organizations standardizing on that language need the custom-priced Enterprise edition. Native vector search combines graph relationships with unstructured data, making NebulaGraph relevant to AI-oriented data fusion rather than graph-only applications.

Key features

Graph algorithms are included, and official clients for Java, Python, C++, and Go provide established integration paths. Spark and Flink connectors connect graph processing with data-engineering pipelines. NebulaGraph Studio is a web interface for schema creation, data import, nGQL queries, and graph exploration; it gives teams a practical visual companion to programmatic access.

Deployment is flexible across on-premises infrastructure, public clouds, and hybrid environments, including macOS and Windows setups. NebulaGraph Cloud can place clusters on Microsoft Azure, AWS, or Google Cloud Platform. Cloud security includes Private Link for private network access and role-based permissions at graph-space level. Raft maintains redundant backups across server rooms in the same city, while snapshots support point-in-time restoration and automatic replacement of failed instances. Cloud customers can open dashboard work orders for technical assistance.

There is one important data-model caveat: ACID transactions are not supported. Applications requiring transactional guarantees should therefore evaluate their consistency requirements before adopting NebulaGraph.

Pricing

Open-source Edition — free. This Apache 2.0 release provides a subset of core features with community collaboration and support. It is the natural starting point for self-managed projects and teams willing to operate their own infrastructure, but it does not include the broader enterprise feature set.

NebulaGraph Cloud Standard — priced on request, billed pay-as-you-go. A single node can reach 8 vCPUs, 64 GB RAM, and 320 GB storage. Private endpoints and IP allow lists improve network control, but there is no high availability, making Standard better for less critical workloads than services that require failover.

NebulaGraph Cloud Professional — priced on request, billed pay-as-you-go. It supports up to 10 query nodes and 10 storage nodes, adds high availability, and retains private endpoint and IP allow-list controls. Faster-response technical support makes this the stronger cloud choice for production systems where interruption is costly.

Enterprise Edition — custom pricing. Enhanced performance, enterprise-grade stability and security, premium support, and native ISO-GQL support in v5.0 target organizations with demanding operational or language-standard requirements.

The free NebulaGraph Cloud trial lasts 14 days, requires no payment information, and provides one database instance with 2 vCPUs, 4 GB RAM, and 10 GB storage. That is enough for evaluation, not a representation of production capacity. Included storage is 10 GB, so storage-heavy trials will reach that boundary quickly.

Platforms

NebulaGraph supports API, Linux, macOS, self-hosted, web, and Windows environments. This range covers local development, managed browser tooling, and independently operated deployments. Cloud clusters run on Azure, AWS, and Google Cloud Platform, allowing teams to align hosting with an existing provider.

Who it's for

NebulaGraph is a good fit for engineering teams building graph-centric products, especially recommendation, fraud, social, security, knowledge, network, or supply-chain applications. It is also suited to organizations combining graph and unstructured data for AI workflows, or those needing a choice between self-managed and managed-cloud operation.

Choose the open-source edition when Apache 2.0 licensing and community support are sufficient. Choose Standard for a managed single-node deployment, and Professional when high availability, multiple query and storage nodes, and faster support justify cloud spending. Enterprise is the appropriate route for ISO-GQL, enhanced performance, and premium operational support. Systems that depend on ACID transactions should look elsewhere.

Pros and cons

Pro: Trillion-scale graph capacity is aimed at unusually large connected datasets.

Pro: Self-hosted, hybrid, and three major public-cloud options give deployment teams meaningful control.

Pro: Native vector search, graph algorithms, language clients, and Spark/Flink connectors cover both application and data-pipeline needs.

Pro: Professional cloud adds high availability, while snapshots and automatic failed-instance replacement strengthen recovery.

Con: ACID transactions are unavailable, ruling it out for workloads that require transactional semantics.

Con: Standard cloud has no high availability and is capped to a single node, so critical workloads need Professional or Enterprise.

Con: ISO-GQL support is limited to Enterprise v5.0, creating a substantial edition requirement for standards-led teams.

Alternatives

Graph Databases is the broader category guide for comparing graph products. Memgraph is a sensible alternative when an open-source, in-memory database with ACID transactions and on-disk persistence is more important than NebulaGraph’s deployment breadth. Sparksee suits academic users who qualify for its free academic licence and need mobile or desktop platform coverage. LadybugDB is worth considering for an MIT-licensed, free database with pre-compiled binaries and no trial requirement. ArcadeDB offers a free Community edition with its full feature set for teams preferring that distribution model. Aerospike Database is an option for teams evaluating a commercial database with a community tier. OrientDB provides a free community distribution with paid support. TigerGraph is an alternative with a free cloud trial. AllegroGraph offers a free tier capped at 5 million triples.

Verdict

NebulaGraph is a strong choice for teams that need a scalable graph engine, vector search, and freedom to run on-premises, in a hybrid setup, or across major clouds. Start with the open-source edition for self-managed development, or choose Professional when cloud high availability and support are essential. Look elsewhere if ACID transactions are mandatory, if single-node Standard limits are too restrictive, or if ISO-GQL must be available without Enterprise pricing.

Verdict

NebulaGraph offers a free open-source edition and graph database tooling for several deployment models. Teams should check edition-specific query support and Cloud availability needs before selecting a plan.

Where Specifiction places it

Alternatives to NebulaGraphAll alternatives →