ArangoDB vs Alopex DB

What ArangoDB is

ArangoDB is a native multi-model database for key-values, documents, and graphs. One query language, AQL, works with graphs, JSON documents, search, vectors, and geospatial data.

How it works

Key-value: every document has a _key attribute, which is its primary key. Every collection has an index on _key.
Documents: JSON documents are searched and aggregated at the attribute level, and queries can span multiple collections.
Graphs: nodes and edges are both documents. Edges reference nodes through _from and _to and live in edge collections. Traversals follow edges outbound, inbound, or in any direction.
Vector indexes: implemented with the Faiss library and queried with AQL vector similarity functions.

Main features

AQL gives multi-hop graph traversal, shortest paths, full-text search, geospatial search, and vector search. Named graphs keep the graph consistent: removing a node also removes its connected edges.

Limits

From 3.12, the source code and the Community Edition moved from Apache 2.0 to the source-available Business Source License 1.1.
Vector indexes require starting the server with the --vector-index option. Once enabled, it cannot be disabled.
Up to 3.12.8, the embeddings must already be in the documents before a vector index is created.
Graph operations without named graphs do not enforce graph consistency. Even with named graphs, low-level operations can leave dangling edges.
Edges are always directed and cannot point both ways.

How it differs from Alopex DB

Both handle several data models and vector search in one database.

Alopex DB is an embedded database queried with SQL. Data lives in a single .alopex file, and table rows and vectors are updated in the same transaction. The source code is published under Apache-2.0.

Alopex DB is a multi-model database with tables, KV and vectors in one file. Give it a try from the quickstart.

References

ArangoDB: Data Models, ArangoDB documentation, Vector indexes, Incompatible changes in ArangoDB 3.12