Native lakehouse engine for Delta Lake and Apache Iceberg
Website · Docs · Install · Pricing
Most lakehouse jobs don't need a cluster. They need a native engine.
DeltaForge reads and writes Delta Lake and Apache Iceberg in one native process, on your hardware. Serve BI directly from the lake, run graph workloads on the same tables, and give AI agents typed catalog access. No second warehouse, no copy pipeline, no cluster tax.
- Lakehouse SQL Read and write Delta Lake and Apache Iceberg in one engine, in place.
- BI drivers ODBC and ADBC connect Power BI, Tableau, and Excel straight to your tables, with no duplicate copy to license or keep in sync.
- Graph Run Cypher over your Delta tables as native property graphs, on the same data.
- AI agents A Model Context Protocol server gives Claude, Cursor, and Copilot typed catalog access.
5x to 8x faster than Apache Spark on standard reads, and about 4x faster writing 10 million rows to plain Delta. 7,137 bi-directional conformance scenarios are verified against Apache Spark, with results open for inspection.
DeltaForge ships through the native package manager for each platform: Homebrew,
Scoop, winget, and Chocolatey, plus an apt repository, container images on GHCR, the
deltaforge-adbc Python package, and the deltaforge-mcp npm package. Current
instructions: deltaforge.org/install.
- delta-forge Engine, desktop app, and CLI releases
- delta-forge-odbc ODBC driver for Power BI, Tableau, Excel, and DBeaver
- delta-forge-adbc ADBC driver and the Power BI connector
- delta-forge-mcp Model Context Protocol server
- apt-repo Debian and Ubuntu packages
DeltaForge is customer-installed: run it on cloud VMs, on-prem, or air-gapped. It is commercial software with a free Community license. See deltaforge.org/pricing.