TuringDB

TuringDB is an OLAP in-memory graph database engine for critical industries. Built in C++.
Star the repo

[TuringDB @TuringDB_ai]

LingoDB paper on query compilation using MLIR, very interesting ideas https://t.co/PoECoC5ywp

[TuringDB @TuringDB_ai]

We designed TuringDB at first as an analytical graph DB focused on read-heavy graph workloads but we are actually very good at writes.
We are actually faster than Neo4J as soon as writes are batched reasonably. We were surprised by our own benchmarks.

[TuringDB @TuringDB_ai]

Most graph DBs give you one view: now. Need last quarter's graph? Hope you kept a backup.

TuringDB versions graph state like Git versions code. Branch, merge, roll back, query any past version at full speed.

Read: https://t.co/0RFdgGcAG5
Play: https://t.co/ztgNfqWFI3

[TuringDB @TuringDB_ai]

298× faster means:

1 hour of graph computation → ~12 seconds
30 minutes → ~6 seconds
10 minutes → ~2 seconds

This is the difference between offline, batch-only analysis, and interactive or near–real-time decision-making.

GitHub repo: https://t.co/Zs7nBCrGaz

[TuringDB @TuringDB_ai]

One of the most requested features in TuringDB history just shipped: in-process mode.

Embed a full graph database directly inside your Python script. No server to launch, no network round-trips, no infrastructure to manage.

Query your graph in three lines of code.

[TuringDB @TuringDB_ai]

We are working on query compilation for Cypher in TuringDB based on some new, secret for now, SSA IR. It's all very exciting.

[TuringDB @TuringDB_ai]

You can now easily build your graph applications from knowledge graphs, context graph, graph based memory or #graphRAG simply using TuringDB Skills with Claude Code.
People have used it for biomedical research, supply chain, manufacturing, finance, AI agents and more
#claudecode