ThemisDB
A high-performance C++ multi-model database combining graph, vector, relational and file storage under one query language with MVCC, aimed at applications that need native AI and LLM data integration.
ThemisDB is a hybrid database engine written in C++ that unifies graph, vector, relational and file storage behind a single query language, AQL, with multi-version concurrency control, built on RocksDB for its storage layer.
The project tracks module-level readiness explicitly: as of its current snapshot, 19 of 72 modules are marked production-candidate, 48 are still hardening, 2 are experimental, including its llama.cpp and stable-diffusion integrations, and 3 are thin placeholders, with the graph module specifically reported as production-candidate with zero known gaps from its latest verification pass.
The README discloses that the codebase was partially or fully generated with AI-assisted coding tools, and asks users to review it for cross-module consistency, subtle bugs and outdated practices before relying on it in production.