Galactus DB vs Neo4j.
Galactus DB speaks Cypher and Bolt, so Neo4j applications and drivers can often move across with few changes. Here is where the two differ, so you can tell which one fits.
| Galactus DB | Neo4j | |
|---|---|---|
| Query language | Cypher, with documented coverage | Cypher |
| Protocol | Bolt 4.4 and 5.0–5.4; TCP, TLS and WebSocket on one port | Bolt |
| Drivers | Official Neo4j drivers, plus native drivers for C#, Rust, Java, Go, Node, C++ and Python | Official Neo4j drivers |
| Graph algorithms | Built in: centrality, communities, similarity, FastRP, pathfinding | Graph Data Science library, installed as a plugin |
| APOC | Selected functions and procedures built in | APOC library, installed as a plugin |
| Indexes | Range, hash, composite, full-text, vector, point and spatial shape | Range, text, full-text, vector and point |
| Deployment | Single node. The database image is about 8 MB | Single node in Community Edition; clustering in Enterprise Edition |
| Writes | Reads run together; one write executes at a time | Concurrent write transactions |
| Licence | Commercial, with free Developer and Personal editions | Community Edition under GPLv3; Enterprise Edition under a commercial licence |
| Moving data in | Reads Neo4j 4.x and 5.x .dump files directly | — |
Measured side by side.
Full resultsWe ran the same graph and the same queries against Galactus DB and Neo4j 5.26.30 Community on 17 September 2026: 100,000 people, 200,000 relationships and an index on the person ID, sent over Bolt by the same Python client. Each container had 8 CPUs and 8 GB of RAM.
| Workload | Galactus DB | Neo4j |
|---|---|---|
| Bulk load, nodes per second | 67,976 (53,302 in group mode) | 36,297 |
| Index seek, p50 | 0.49 ms | 1.07 ms |
| Two-hop neighbourhood count, p50 | 0.50 ms | 1.07 ms |
| 64 concurrent readers, seeks per second | 3,904 | 3,016 |
| Single client, durable autocommit, p50 | 4.79 ms (group mode) | 3.88 ms |
| Peak memory during 16-client writes | 402 MB | 2,793 MB |
Galactus DB led on batch loading, indexed lookups, short traversals and concurrent reads. Neo4j had the lower latency for a single client making individual durable commits. Neo4j's memory reflects its configured 2 GB heap and 2 GB page cache, and neither figure is a minimum requirement. The test did not cover graphs larger than memory, clustering or Neo4j Enterprise Edition.
Choose Galactus DB when…
- You want Cypher and your Neo4j drivers on a database that is small to run.
- Bulk loads, indexed lookups and short traversals dominate the workload.
- You want graph algorithms and APOC utilities without managing plugins.
- The graph's structure and indexes fit in memory on one server.
Choose Neo4j when…
- You need a replicated cluster or automatic failover.
- The graph is much larger than the memory you can give it.
- Many clients write at once and need parallel write execution.
- You need label- and property-level security, a persistent GDS graph catalogue or ML pipelines.
Moving from Neo4j.
- Create a dump from a cleanly shut down Neo4j 4.x or 5.x database. Dump import guide
- Import it into an empty database from Explorer, or with
CALL gdb.neo4j.import. - Recreate your indexes and constraints. Indexes and constraints
- Point your driver at
bolt://your-host:7687. Start the server withGDB_NEO4J_AGENT=truefor the Neo4j Python driver. Connecting over Bolt - Run your queries and check anything unusual against the compatibility guide.
Questions
Is Galactus DB a drop-in replacement for Neo4j?
For many applications it is close. Galactus DB speaks Cypher over Bolt 4.4 and 5.0–5.4, accepts the official Neo4j drivers and includes graph algorithms and selected APOC procedures. It is not a drop-in replacement for everything: check your queries against the compatibility guide, and recreate indexes and constraints after moving your data.
Can I use the Neo4j drivers with Galactus DB?
Yes. The official Neo4j drivers connect with bolt://, bolt+s:// and neo4j:// URLs. The Python driver only accepts servers that identify as Neo4j, so start Galactus DB with GDB_NEO4J_AGENT=true when you use it. Native Galactus DB drivers are also available for C#, Rust, Java, Go, Node, C++ and Python.
Can I import an existing Neo4j database?
Yes. Galactus DB reads offline Neo4j 4.x and 5.x .dump files directly, from Explorer or with CALL gdb.neo4j.import. Nodes, labels, properties and relationships are imported in one atomic statement. Schema, indexes, constraints and users are not imported, so recreate those afterwards.
Does Galactus DB support clustering?
No. Galactus DB runs on a single node. Routing drivers are given a single-node routing table, so they connect, but there is no replication or failover. If you need a cluster today, Neo4j Enterprise Edition provides one.
How is Galactus DB licensed?
Galactus DB is commercial software with free Developer and Personal editions. A fresh instance starts in Developer mode with no account or licence key. Self-hosted business use needs a Commercial licence at $400 per core, per instance, per year, with discounts for longer prepaid terms.
Neo4j is a registered trademark of Neo4j, Inc. Galactus DB is not affiliated with Neo4j, Inc. Neo4j details are from its public documentation; check the current Neo4j documentation for your version.
Measure it on your own workload.
The Developer edition is free and needs no account or licence key. Bring your queries, or a Neo4j dump, and see how they run.