2025 was the year ArcadeDB shifted into high gear. We started the year by modernizing our foundation and ended it by pushing the boundaries of vector search and remote connectivity.
Major Features & Innovations
Next-Gen Vector Search with JVector (v25.11.1 - v25.12.1)
The AI revolution requires high-performance vector indexing. In late 2025, we integrated the JVector engine, introducing a new LSMVector index.
-
Filtered Vector Search: You can now perform precise queries by combining vector similarity with traditional metadata filters.
-
Quantization Fixes: We resolved critical bugs in INT8 and BINARY quantization, ensuring accuracy across all dimensions.
-
Compression: Enhanced compression for LSM vector indexes significantly reduces the disk footprint for large-scale embeddings.
Python Embedded Bindings (v25.10.1)
We expanded our ecosystem significantly by releasing Official Python Bindings. For the first time, Python developers can run ArcadeDB in embedded mode directly within their applications, eliminating network overhead for local, high-performance graph and document management.
The gRPC Protocol & OpenAPI (v25.9.1)
To complement our existing HTTP/JSON and binary protocols, we introduced experimental gRPC support. This provides a high-performance, strongly-typed alternative for server communication. Additionally, we added an OpenAPI 3.0 specification endpoint (/api/v1/openapi.json), making API discovery and client generation easier than ever.
Storage Engine Evolution: RID Recycling (v25.3.1)
Early in the year, we implemented a major change in the storage engine: full reuse of deleted space and recycling of Record IDs (RIDs). This ensures that long-running databases remain lean and efficient without requiring manual export/import cycles.
Advanced SQL & Schema Capabilities
-
Embedded List Indexing: You can now index properties inside embedded documents or maps within lists using the
BY ITEMsyntax. -
Dynamic Type Names: SQL commands now support placeholders for type names (e.g.,
INSERT INTO $type), enabling more reusable code. -
Type Aliasing: Support for
ALTER TYPE ... NAMEand aliases allows for seamless schema evolution. -
Unidirectional Edges: Added the ability to define edges that only point one way, optimizing storage for specific graph use cases.
Improvements & Performance
-
Java 21 Migration (v25.4.1): We moved our core to Java 21, leveraging modern language features and the latest JVM performance optimizations.
-
High Availability (HA) Enhancements: Improved connection strategies for HA clusters, including support for HTTP proxies and better handling of replica resynchronization.
-
Cloud Native: Initial Helm chart implementation and a migration to Temurin-alpine for Docker images to ensure a smaller, more secure footprint.
-
Encrypted Backups (v25.5.1): Added native support for encrypted backup and restore to ensure data security at rest.
-
Query Optimization: Massive improvements to index awareness for
CONTAINS,IN, andCONTAINSALLoperators, and a newifempty()SQL function.
Bug Fixes & Stability
Our community helped us squash hundreds of bugs this year. Key highlights include:
-
Data Integrity: Fixed critical issues where unique constraints were bypassed during
UPDATEoperations. -
Driver Compatibility: Major stability fixes for the PostgreSQL driver, resolving CPU spikes on idle connections and improving Cypher parametric query support over JDBC.
-
Consistency: Resolved “Concurrent modification on page” errors and fixed issues where
ORDER BYwould occasionally return duplicate or incorrect results when using indexes. -
Studio Enhancements: Fixed various UI issues in ArcadeDB Studio, including login errors and notification displays.
Security & Dependencies
We maintained a rigorous update schedule, keeping our core libraries up to date to mitigate vulnerabilities (like the commons-lang3 and commons-beanutils fixes).
-
Lucene: Upgraded to 10.3.2.
-
Netty & gRPC: Continuous updates to the latest stable versions.
The Community
This year, we welcomed numerous new contributors and expanded our integrations.
As we move into 2026, our focus remains on making ArcadeDB faster and rock-solid.
Thank you for being part of this journey!
Ready to try the latest features? Download ArcadeDB v25.12.1 or check out the Full Changelog on GitHub.