GraphForge vs. NeuG
Two young embedded Cypher engines with different priorities. Which fits your workload?
NeuG is Alibaba's Apache-2.0 graph database, open-sourced in February 2026. It reuses GraphScope Flex's C++ storage and query engine behind a DuckDB-style embedded interface, and the same runtime can serve network clients through db.serve(). Like GraphForge, it runs Cypher in-process and adds algorithms, vector search, and full-text search. It leans toward transactional serving; GraphForge leans toward analysis and investigation history.
On this page
Should I choose NeuG or GraphForge?
NeuG when one runtime must serve embedded and as a service; GraphForge for algorithm breadth and investigation history.
Choose GraphForge when
- You need a broad algorithm catalog (94 algorithms) beyond NeuG's GDS extension.
- Your analysis needs checkpoints, provenance, and recorded competing hypotheses.
- You want results as Arrow tables and a project persisted as Parquet.
Choose NeuG when
- One runtime must work embedded during development and as a concurrent service in production.
- You need HNSW approximate vector search committed atomically with graph changes.
- You want Java or C++ bindings, or the GraphScope team's engine lineage.
What can each one do?
21 of 26 capabilities differ. For 15, NeuG's docs did not confirm support, so they are marked Not established rather than No.
Each capability has a status and a source. “Not established” means we could not confirm the feature from NeuG's primary documentation. It does not mean NeuG lacks it.
| Capability | GraphForge | NeuG |
|---|---|---|
| Storage | ||
| Property graph model | Shipped Labeled property graph with typed properties. 1 | Yes Graph-native database queried with Cypher over labeled nodes and relationships. 17 |
| Document model | Not offered No document-store model; properties accept nested, heterogeneous values. 1 | Not established JSON/JSONL import exists; no document model documented. |
| Relational tables | Not offered Not a relational database; results arrive as Arrow tables. 1 | Not established No SQL interface documented. |
| Key-value access | Not offered No key-value API; data is a property graph. 1 | Not established No key-value API documented. |
| Columnar storage | Shipped Durable projects store graph data as Arrow and Parquet. 4 | Not established Storage layout not described in primary docs; engine reused from GraphScope Flex. |
| Vector search | Shipped Exact cosine retrieval over embeddings you supply; no ANN index. 5 | Extension vector_search extension (v0.2.0): HNSW with l2, cosine, inner-product; built on zvec. 2717 |
| Full-text search | Shipped Tantivy-backed BM25; hybrid text and vector retrieval through find(). 5 | Extension fts extension (v0.2.0): BM25 over SQLite FTS5 indexes, Jieba tokenizer for Chinese. 2725 |
| Geospatial | Not offered No spatial types or indexes are documented. | Not established No spatial types or extension listed. |
| Execution | ||
| Transactional workloads (OLTP) | Partial Local analytical scope; durable atomic writes with queued and optimistic writer modes. 6 | Yes Service (TP) mode uses MVCC snapshots for concurrent sessions; updates serialized via an update lease. 22 |
| Analytical workloads (OLAP) | Shipped Analytical verbs and Cypher over a local columnar project. 7 | Yes Embedded (AP) mode targets local analytics; graph algorithms run in place without exporting. 2220 |
| Graph traversal | Shipped Cypher MATCH patterns plus BFS, DFS, Dijkstra, A*, and reachability. 8 | Yes Cypher MATCH pattern matching, plus a Pattern Match extension for subgraph matching. 2625 |
| Variable-length paths | Shipped Bounded and unbounded patterns and shortestPath(). 8 | Yes [*min..max] with TRAIL, ACYCLIC and unweighted SHORTEST options. 26 |
| Graph algorithms | Shipped 94 native algorithms behind analysis verbs, not Cypher CALL procedures. 37 | Extension GDS extension (since v0.1.3): PageRank, WCC, Louvain, Leiden, K-Core, etc., over projected graphs. 21 |
| Incremental computation | Not offered Algorithms recompute on request; no incremental mode is documented. 7 | Not established No incremental algorithm execution documented; indexes are maintained incrementally on writes. |
| Deployment | ||
| Embedded / in-process | Shipped In-process library; no server, daemon, or connection string. 9 | Yes In-process library (e.g. import neug in Python). 1719 |
| Server mode | Planned Optional remote peer extension planned after 0.6.0. 11 | Yes db.serve() exposes the same runtime as a network service for concurrent clients. 17 |
| WebAssembly | Not planned A browser WebAssembly engine was declined in ADR 0027. 12 | Not established No WebAssembly build documented. |
| Distributed cluster | Not planned Distributed clusters and HA replication are out of scope. 9 | Not established No clustering or sharding documented for NeuG itself. |
| Managed cloud | Not planned No hosted compute. The graphforge.sh Hub distributes repositories; execution stays local. 13 | Not established No hosted offering found. |
| Investigation | ||
| Branching | Experimental Research versions, branches, slices, proposals, and forks ship in 0.6.0 release candidates. 14 | Not established No branching feature documented. |
| Snapshots / time travel | Shipped Named checkpoints you can diff and revert, plus transaction-time snapshots. 3 | Not established No as-of queries documented; read-only transactions pin a consistent view only. |
| Immutable history | Shipped Knowledge ledger records immutable assertions and evidence lineage. 16 | Not established No immutable version history documented. |
| Provenance | Shipped Provenance and knowledge ledger with evidence lineage. 16 | Not established No provenance tracking documented. |
| Competing hypotheses | Shipped Immutable hypothesis groups with membership and selection events. 3 | Not established |
| Operations | ||
| ACID transactions | Partial Acknowledged-durable atomic publication; docs do not claim general ACID or serializable isolation. 6 | Yes WAL plus checkpoints; explicit transactions (v0.2.0); isolation level not named in docs. 2227 |
| Replication | Not planned HA replication is out of scope. 9 | Not established No replication documented. |
Showing 26 of 26 capabilities.
How is each one built?
NeuG: transactional property graph. GraphForge: an embedded property graph library in Rust, persisted as Parquet.
GraphForge
GraphForge is a Rust library that runs inside your process: no daemon, server, or connection string. Projects persist to a local directory as Arrow and Parquet data plus JSON metadata, and every query and analysis verb returns an Apache Arrow table. Algorithms run in the same native engine as Cypher and are called as API verbs (rank, cluster, paths, analyze, similar), not as Cypher procedures.
NeuG
NeuG embeds the GraphScope Flex storage and query engine, compiles Cypher with a compiler adapted from Kuzu, and loads extensions in the style of DuckDB. Embedded mode admits one writer at a time. Service mode uses MVCC snapshots and serializes updates through a lease. Vector and full-text indexes commit with graph changes and recover through the WAL.
NeuG and GraphForge make similar embedding choices. NeuG adds a built-in service mode and approximate vector indexes; GraphForge adds a much larger algorithm library and investigation history.
What the documentation says about NeuG
- The GraphScope team describes NeuG as a lightweight embedded graph database that reuses GraphScope Flex's storage and query engine with a new interface layer.19
- The team cites DuckDB's embeddable, serverless experience as the model it wanted for graph data.19
- One runtime runs in-process or as a network service via db.serve().17
- Embedded mode uses exclusive write admission, while service mode uses MVCC snapshots and serializes updates through an update lease.22
- NeuG's C++ Cypher compiler is adapted from Kuzu, and its extension framework is inspired by DuckDB.17
- Vector and full-text index changes commit atomically with graph changes and recover through checkpoints and the WAL.23
- v0.2.0 (2026-09-03) added HNSW vector search, BM25 full-text search, graph namespaces and explicit transactions.27
- v0.2.1 (2026-09-30) added native Windows wheels for x86_64 and ARM64 and named path aliases.18
- The query layer is built on GOpt's unified intermediate representation, designed with future ISO GQL compatibility in mind.19
Which one is faster?
Not measured yet. No head-to-head run has been published, so this page shows no performance chart.
Charts appear only after a reproducible run exists: pinned versions, disclosed hardware, generated data, raw results, and scripts.
No comparative results yet
The suites below apply to GraphForge and NeuG. Each will be reported separately per benchmark class, never merged into one leaderboard.
Ingestion and writes
| Suite | Workloads | Metrics | Status |
|---|---|---|---|
| INGEST | CSV, JSONL, Parquet, graph edge lists | Records/s, MB/s, total elapsed | Not run |
| MUTATE | Insert, update, delete, upsert | Ops/s, p50/p95/p99 latency | Not run |
Query, analytics, and system
| Suite | Workloads | Metrics | Status |
|---|---|---|---|
| LOOKUP | ID lookup, indexed property filter | p50/p95/p99 latency | Not run |
| TRAVERSE | BFS, k-hop, variable-length path | Latency, visited edges/s | Not run |
| ALGO | PageRank, WCC, SCC, Louvain, SSSP | Time, memory, correctness | Not run |
| HYBRID | Vector + graph + metadata filters | Recall@k, latency, throughput | Not run |
| EMBED | Startup, load, reopen, binary size | ms, MiB, disk footprint | Not run |
| CONCUR | Mixed readers and writers | Throughput, tail latency | Not run |
Which graph algorithms ship?
GraphForge: 94 native algorithms. NeuG: 10 named in its docs, delivered as extension.
Algorithms are listed by name, as documented. Counts of extension modules or third-party libraries are not presented as native coverage.
GraphForge
94 Rust algorithms run in the engine behind five analysis verbs (rank, cluster, paths, analyze, similar) and return Arrow tables.37
A selection of the catalog:
- PageRank
- ArticleRank
- Betweenness (Brandes)
- Closeness
- Harmonic closeness
- Eigenvector
- HITS
- CELF influence maximization
- Triangle count
- K-core decomposition
- Louvain
- Leiden
- Label propagation
- Infomap
- Walktrap
- Girvan-Newman
- Weakly connected components
- Strongly connected components
- Dijkstra
- A*
- Bellman-Ford
- Yen's k-shortest paths
- Max flow / min cut
- Steiner tree
- Minimum spanning tree
- Topological sort
- Node2Vec
- GraphSAGE
- Fast random projection
- Node similarity (Jaccard)
NeuG
Graph algorithms ship in the official GDS extension (LOAD gds), invoked with CALL over a projected-graph namespace; unweighted shortest path is also native Cypher syntax.2126
Delivered as: extension
- PageRank
- BFS
- Single-Source Shortest Path (SSSP)
- Weakly Connected Components
- Local Clustering Coefficient
- K-Core Decomposition
- Label Propagation (CDLP)
- Louvain
- Leiden
- Shortest path (SHORTEST)
Where can each one run?
GraphForge runs in-process only. NeuG: embedded yes, server yes, cluster not established.
| Mode | GraphForge | NeuG |
|---|---|---|
| Embedded / in-process | Shipped In-process library; no server, daemon, or connection string. 9 | Yes In-process library (e.g. import neug in Python). 1719 |
| Server mode | Planned Optional remote peer extension planned after 0.6.0. 11 | Yes db.serve() exposes the same runtime as a network service for concurrent clients. 17 |
| WebAssembly | Not planned A browser WebAssembly engine was declined in ADR 0027. 12 | Not established No WebAssembly build documented. |
| Distributed cluster | Not planned Distributed clusters and HA replication are out of scope. 9 | Not established No clustering or sharding documented for NeuG itself. |
| Managed cloud | Not planned No hosted compute. The graphforge.sh Hub distributes repositories; execution stays local. 13 | Not established No hosted offering found. |
| Replication | Not planned HA replication is out of scope. 9 | Not established No replication documented. |
What will my team write?
GraphForge: openCypher from Python, Node.js, Rust. NeuG: Cypher.
NeuG exposes Cypher through Python, Node.js, Java, C++, and a CLI, with algorithms, vector search, and full-text search loaded as extensions and called with CALL. GraphForge exposes them as Python, Node.js, or Rust verbs with no extension loading.
- Query languages
- GraphForge openCypher
- NeuG Cypher
- SDKs and bindings
- GraphForge Python, Node.js, Rust
- NeuG Python, Node.js, Java, C++, CLI
- Written in
- GraphForge Rust
- NeuG C++
- Surfaces
- GraphForge Python, Node.js, Rust crates, CLI, VS Code extension, Notebooks, Agent skills
What are the trade-offs?
NeuG is young, with narrow index support and single-writer embedded mode. GraphForge lacks a service mode and HNSW.
A fair assessment cuts both ways.
GraphForge
- No built-in service mode; a remote peer extension is planned after 0.6.0.
- Vector search is exact rather than HNSW.
- No Java or C++ bindings.
NeuG
- Bulk LOAD FROM and COPY run only in Embedded mode; Service mode rejects them.22
- Most GDS algorithms require a homogeneous projected subgraph; only Label Propagation, Louvain and Leiden accept multiple labels.21
- Embedded mode is effectively single-writer: an explicit read-write transaction holds exclusive write admission until it ends.22
- Index support is still narrow: only HNSW and FTS exist, with BTree scalar and composite indexes listed as planned.23
How would I migrate?
Load NeuG data through Arrow, pandas, or Polars. Standard openCypher carries over; procedure calls become verb calls.
Into GraphForge
- GraphForge 0.5.x loads data through Arrow, pandas, and Polars bulk calls or Cypher scripts; it has no native CSV loader.15
- Algorithms are API verbs (rank, cluster, paths, analyze, similar), not Cypher CALL procedures, so procedure calls become verb calls.7
- GraphForge passes the full openCypher 2024.3 TCK, so standard openCypher queries are the portable part of a migration.10
Moving from NeuG
What else do people ask about NeuG?
What is NeuG? NeuG is Alibaba's Apache-2.0 embeddable graph database built on GraphScope Flex, with a GDS extension.
What is NeuG?
NeuG is an embeddable graph database from Alibaba's GraphScope team, released under Apache 2.0. It reuses GraphScope Flex's engine and runs in-process or as a network service.
How is NeuG related to GraphScope?
NeuG's README names GraphScope Flex as the basis of its storage and query engine, and the GraphScope homepage describes NeuG as powered by GraphScope technology. GraphScope itself is a distributed graph computing platform.
Which algorithms does NeuG ship?
Its GDS extension provides PageRank, BFS, single-source shortest path, weakly connected components, local clustering coefficient, k-core, label propagation, Louvain, and Leiden. Most require a homogeneous projected subgraph.
How was this researched?
From 27 primary sources, accessed 2026-10-19, each cited where it is used.
Capabilities were checked against primary documentation, repositories, and release notes, accessed 2026-10-19. Each cell records two things: the capability’s state and the evidence behind it. Every GraphForge cell carries a release status. Shipped means stable 0.5.2; experimental means 0.6.0 release candidates only. GraphForge is pre-1.0 and makes no backward-compatibility guarantee yet.
- Yes documented natively
- Partial limited or caveated
- Extension official extension or module
- Paid edition paid or enterprise edition only
- Removed removed, with the version cited
- Not established not confirmed by our research; not a “no”
- Shipped in stable 0.5.2
- Partial shipped with documented limits
- Experimental in 0.6.0 release candidates only
- Planned on the roadmap
- Not offered outside the product's model
- Not planned explicitly out of scope
Benchmark classes
- Equivalent workloads. The same logical operation with the same correctness requirement on every engine.
- Native optimized workloads. Each engine's best supported implementation, including projections, indexes, extensions, and native algorithms.
- End-to-end application workloads. Time from raw data to a usable result: import, indexing, projection, computation, and output retrieval.
Sources · accessed 2026-10-19
- GraphForge repository README — GitHub
- GraphForge releases — GitHub
- GraphForge 0.5.0 changelog — GitHub
- Storage architecture — GraphForge Book
- Analytics integration: search and retrieval — GraphForge Guide
- ADR 0018: acknowledged durability and isolation — GitHub
- Algorithm verbs — GraphForge Book
- Cypher guide — GraphForge Guide
- openCypher compatibility — GraphForge Reference
- openCypher TCK compliance — GraphForge Reference
- Remote authority peer extension (#215) — GitHub
- ADR 0027: native runtime boundary — GitHub
- Release scope and roadmap — GraphForge Docs
- Keep research history — GraphForge Guide
- Graph construction — GraphForge Guide
- Knowledge public API v1 — GraphForge Book
- alibaba/neug README — GitHub
- Release v0.2.1 — alibaba/neug on GitHub
- Opensourcing NeuG — GraphScope blog
- Introduction — NeuG docs
- GDS extension — NeuG docs
- Transaction model — NeuG docs
- Storage indexes — NeuG docs
- Import data — NeuG docs
- Extensions overview (doc source) — alibaba/neug on GitHub
- MATCH clause (doc source) — alibaba/neug on GitHub
- Release v0.2.0 — alibaba/neug on GitHub
Try GraphForge on your own graph.
Install the package, open a project, and run the same notebook we measure with.