News & Updates
The latest announcements, events, and product updates from GizmoData
Python ADBC Driver v1.2.0: Connection Profiles
adbc-driver-gizmosql v1.2.0 adds support for ADBC connection profiles — reusable TOML files that bundle your GizmoSQL server URI and connection options, so your application code stays clean and credential-free.
Reference a profile by name — connect(profile="gizmosql_dev") or connect("profile://gizmosql_dev") — or by absolute path. Secrets stay out of the file with {{ env_var(...) }} substitution at connect time, and explicit connection parameters always override profile settings, so one-off overrides don't require editing the profile.
Connection profiles are a cross-language ADBC standard (requires adbc-driver-manager 1.11.0+), so the same TOML file that drives your Python code works from any ADBC-capable runtime. For a hands-on walkthrough using GizmoSQL, see Columnar's cookbook tutorial linked below.
Key Features
Reusable TOML Profiles
Bundle the server URI, username, and driver options in one file. Point your code at a profile name instead of hardcoding connection details.
Credential-Free Code
Secrets are substituted from the environment at connect time via {{ env_var(...) }} — nothing sensitive lands in the profile file or your source.
One-Line Environment Switching
Same client code, different profile name: gizmosql_dev for development, gizmosql_prod for production. No code changes to retarget.
Cross-Language ADBC Standard
Profiles follow the ADBC spec (adbc-driver-manager 1.11.0+), so the same file works across ADBC drivers and language runtimes.
Install / Upgrade
pip install --upgrade adbc-driver-gizmosqlProfile File (gizmosql_dev.toml)
profile_version = 1
[Options]
uri = "grpc+tls://gizmosql.example.com:31337"
username = "gizmosql_user"
# Keep secrets out of the file - substituted from the
# environment at connect time
password = "{{ env_var(GIZMOSQL_PASSWORD) }}"Connect by Profile Name (Python)
from adbc_driver_gizmosql import dbapi as gizmosql
with gizmosql.connect(profile="gizmosql_dev") as conn:
with conn.cursor() as cur:
cur.execute("SELECT 1 AS value")
print(cur.fetch_arrow_table())
Join the New GizmoData Community on Slack
GizmoData just launched a community Slack where GizmoSQL users can talk directly with the founder, swap tips on running DuckDB as an Apache Arrow Flight SQL server, get help, and hear about new releases first. Free to join — everyone welcome.
Philip Moore on the A/I Prosperity Podcast
GizmoData founder Philip Moore joined Torben Andersen on the A/I Prosperity podcast for a wide-ranging conversation on GizmoSQL, running DuckDB as an enterprise-grade Apache Arrow Flight SQL server, and cutting cloud data-warehouse costs.
GizmoSQL Is Now a Built-In DBeaver Driver
As of the 26.1.0 release, GizmoSQL ships as a native, built-in driver in DBeaver — the world's most popular database GUI, with millions of users. There's no manual JDBC setup, no hunting for a JAR, and no driver configuration: open DBeaver, create a new connection, pick GizmoSQL from the driver list, enter your host, port, and credentials, and start querying. The driver speaks Arrow Flight SQL to your GizmoSQL server, so you get the same high-performance, columnar transport you'd get from any other GizmoSQL client — now from a polished, full-featured SQL IDE. Because the driver is bundled and maintained upstream in DBeaver itself, it stays current automatically with every DBeaver release.
What You Get
Built-In Driver
GizmoSQL is bundled directly in DBeaver — select it from the driver list when creating a connection. No download, no JAR, no classpath.
Zero JDBC Setup
Skip manual driver installation and configuration. Enter your host, port, and credentials in the connection dialog and you're querying.
Arrow Flight SQL Transport
The driver connects over Apache Arrow Flight SQL, giving you GizmoSQL's high-performance columnar data transport straight into DBeaver.
Full SQL IDE
Browse schemas, edit and run SQL, view results, and manage your GizmoSQL data with DBeaver's rich, mature database GUI.
Experimental Quack JDBC Driver
quack-jdbc is a new experimental JDBC driver for DuckDB's Quack remote protocol — the native client-server protocol DuckDB Labs shipped in May 2026, stabilizing with DuckDB v2.0 in September 2026.
Point any JDBC-compatible tool (DBeaver, IntelliJ DataGrip, dbt, Spark) at a jdbc:quack://host:9494 URL and it just works — the driver implements the full DatabaseMetaData surface and decodes DuckDB DataChunk vector encodings (FLAT, CONSTANT, DICTIONARY, SEQUENCE).
Status is alpha; expect breaking changes before v2.0. FSST decoding, nested java.sql.Array/Struct wrapping, and native server-side parameter binding are on the roadmap.
Key Features
jdbc:quack:// URLs
Connect any JDBC tool with a standard URL: jdbc:quack://host[:9494]. Auth token and other options pass via Properties or URL parameters.
JVM Tool Compatible
Works with DBeaver, IntelliJ DataGrip, dbt-duckdb-style workflows, and JVM data-engineering frameworks like Spark out of the box.
Full DatabaseMetaData
Catalog, schema, table, view, column, key, and index metadata — so IDE schema browsers and SQL tools light up immediately.
DataChunk Vector Decoding
Decodes DuckDB DataChunk vector encodings: FLAT, CONSTANT, DICTIONARY, and SEQUENCE. (FSST and native parameter binding are roadmap items.)
Install from Maven Central
<dependency>
<groupId>com.gizmodata</groupId>
<artifactId>quack-jdbc</artifactId>
<version>0.1.0-alpha.1</version>
</dependency>
Experimental ADBC Driver for Quack
adbc-driver-quack is a new experimental Apache Arrow ADBC driver for DuckDB's Quack remote protocol — zero-copy Arrow RecordBatches from the server's columnar engine all the way to your client's Arrow-native runtime.
Run standard ADBC queries and pull results as Arrow tables, or stream large results with bounded memory via fetch_record_batch(). Bulk ingest is supported through Statement.BindStream → APPEND_REQUEST.
The recommended choice when you want zero-copy Arrow end-to-end, or a Quack client from a non-JVM runtime like Python, Go, Rust, or R. Requires a Quack server (DuckDB v1.5.2+ with the Quack extension). Status is alpha (v0.1.0-alpha.1).
Key Features
Zero-Copy Arrow End-to-End
Results arrive as Apache Arrow RecordBatches — no row-by-row materialization, no intermediate copies between the DuckDB server and your client runtime.
Python & Go Clients
pip install adbc-driver-quack for Python, or go get github.com/gizmodata/adbc-driver-quack for Go. The same ADBC surface lights up Rust and R too.
Streaming with Bounded Memory
Pull large result sets one batch at a time via fetch_record_batch() — perfect for analytics pipelines that won’t fit in RAM.
Standard ADBC Bulk Ingest
Load data fast with Statement.BindStream → APPEND_REQUEST. Autocommit by default; explicit transactions when you need them.
Install (Python)
pip install adbc-driver-quackInstall (Go)
go get github.com/gizmodata/adbc-driver-quack@latestExample Usage (Python)
import adbc_driver_quack.dbapi as quack
with quack.connect(
uri="quack://localhost:9494",
db_kwargs={"adbc.quack.token": "my-secret-token"},
) as conn, conn.cursor() as cur:
cur.execute("SELECT 42 AS answer")
table = cur.fetch_arrow_table()
GizmoSQL Now Has a Python Library
The new gizmosql Python library lets you spin up a real GizmoSQL server — same engine as the CLI, same Arrow Flight SQL protocol — directly from Python. The Server() context manager launches a managed subprocess on an auto-picked free port and downloads the matching server binary on first use into ~/.cache/gizmosql/, so there's nothing to install ahead of time. The most common use is a session-scoped pytest fixture: every test gets its own real Flight SQL endpoint with no port collisions across pytest-xdist workers. It's equally at home in notebooks, multi-agent workflows, and quick local demos. Install with 'pip install gizmosql[adbc]' (the [adbc] extra adds the ADBC Flight SQL client so Server.connect() works in-process). Stable and LTS channels are both supported via the channel kwarg. Requires Python 3.10+ and runs on macOS arm64, Linux amd64/arm64, and Windows amd64.
Why a Python Library?
One-Line pip Install
pip install gizmosql[adbc] — the [adbc] extra adds the Flight SQL client so Server.connect() works in-process. Drop it for a smaller install.
Server() Context Manager
A real GizmoSQL server in a managed subprocess on an auto-picked free port. The matching server binary downloads on first use.
pytest Fixture Ready
Drop into conftest.py for a session-scoped fixture — every test gets a real Flight SQL endpoint, no port collisions across pytest-xdist workers.
Stable & LTS Channels
Pass channel="lts" to download the LTS server build instead. Same Python API, same protocol — the channel just controls which DuckDB pin runs underneath.
Introducing the GizmoSQL LTS Channel
GizmoSQL now ships in two parallel release channels — Stable and LTS — that share the same GizmoSQL feature set and only differ in which DuckDB release is bundled. The Stable channel tracks the latest DuckDB minor for every new feature, type, and performance improvement on its normal cadence. The LTS channel tracks the most recent DuckDB LTS for production deployments where the underlying database's stability guarantees matter more than new features. Both channels get every GizmoSQL fix, feature, and quality-of-life improvement at the same time — choosing LTS only changes which DuckDB version is statically linked into the binary. The CLI flags, library API, configuration, authentication, and protocol behavior are identical, and the two builds install side-by-side via Homebrew, Docker, and MSI.
Why LTS?
Tracks DuckDB LTS Releases
Pinned to the most recent DuckDB LTS (e.g. v1.4.4) for production deployments. The pin advances when DuckDB names a new LTS.
Side-by-Side With Stable
LTS binaries, images, and packages carry an _lts / -lts suffix so they coexist with Stable on the same machine, registry, or download index.
Identical GizmoSQL API
Same flags, env vars, library API, authentication, and Arrow Flight SQL protocol. Switching channels is a binary swap and a restart.
Homebrew, Docker & MSI
Install via gizmodata/tap → gizmosql-lts, gizmodata/gizmosql-lts on Docker Hub / GHCR, or the LTS MSI on Windows. iOS remains stable-only.
qgizmosql: New QGIS Plugin for GizmoSQL
qgizmosql is a new QGIS plugin that lets you browse and visualize spatial data from a GizmoSQL server directly in QGIS — no raw DuckDB file required. The plugin is being forked from QDuckDB (Oslandia, GPLv2+) with its DuckDB embedded-file connection layer swapped for the adbc-driver-gizmosql client, which speaks Arrow Flight SQL to a remote (or local) GizmoSQL server. Connect, pick a table with a geometry column, and add it as a QGIS layer streamed via Arrow record batches — with feature requests (filters, bbox, attribute selection) pushed down to GizmoSQL as SQL.
Key Features
Arrow Flight SQL Transport
Connects to GizmoSQL via gRPC + TLS using adbc-driver-gizmosql — no local DuckDB file required. Layers stream as Arrow record batches.
Remote & Multi-User
Unlike QDuckDB, qgizmosql talks to a remote GizmoSQL server, so multiple GIS users can analyze the same live dataset concurrently.
Password or OAuth / SSO
Authenticate with username/password or full browser-based OAuth/SSO flows — including enterprise identity providers.
Server-Side Spatial Engine
GEOMETRY types and 100+ DuckDB spatial functions execute server-side on GizmoSQL. Filters and bboxes are pushed down as SQL.
Install from the official QGIS plugin repository — in QGIS go to Plugins → Manage and Install Plugins…, search for qgizmosql, and click Install.
GizmoSQL Now Available on the Apple App Store
GizmoSQL is now live on the Apple App Store for iPhone and iPad. You can spin up a real Arrow Flight SQL server — powered by DuckDB — right from your pocket. It's ideal for development, experimentation, demos, and learning Arrow Flight SQL on the go. It is not intended for production workloads, but the fact that the same GizmoSQL engine runs on iOS shows just how flexible it is: from cloud VMs and Kubernetes to Windows, macOS, and now iPhone and iPad.
What You Get
Real Arrow Flight SQL Server
Runs the same GizmoSQL engine as desktop and cloud — complete with Arrow Flight SQL endpoints and DuckDB-backed query execution.
Runs on iPhone & iPad
From cloud VMs to Kubernetes to the device in your pocket — GizmoSQL is incredibly flexible about where it runs.
Perfect for Development & Learning
Prototype queries, test client integrations, and explore Arrow Flight SQL without standing up any infrastructure. Not intended for production workloads.
One-Tap Install
Download directly from the Apple App Store — no Docker, no provisioning, no config files. Just launch and start querying.
GizmoSQL Grafana Plugin Now Available on Grafana Marketplace
The GizmoSQL data source plugin for Grafana is now officially listed on the Grafana Marketplace. Connect Grafana directly to GizmoSQL using the Arrow Flight SQL protocol for high-performance observability dashboards powered by DuckDB. Install it from the Grafana catalog with a single click and start building real-time dashboards over your GizmoSQL data.
Key Features
Official Grafana Marketplace Listing
Install directly from the Grafana plugin catalog — no manual builds or sideloading required.
Arrow Flight SQL Protocol
High-throughput, columnar data streaming from GizmoSQL to Grafana using Apache Arrow Flight SQL.
Real-Time Dashboards
Build live dashboards and alerts over your GizmoSQL data with full SQL query support.
Secure Authentication
Supports username/password and token-based authentication for secure access to your data.
GizmoSQL Now Available for Windows x64
GizmoSQL is now available as a native Windows installer (MSI) for x64 systems. Windows users can now run terabyte-scale analytics locally or on Windows Server infrastructure — at a fraction of the cost of cloud data warehouses like Snowflake, Databricks, or BigQuery. The MSI installer provides a seamless setup experience with full support for GizmoSQL's Arrow Flight SQL engine, DuckDB-powered query execution, and all the connectors your BI and AI tools already use.
What's Included
Native Windows Installer
One-click MSI installer for Windows x64. No Docker, no WSL — runs natively on Windows 10/11 and Windows Server.
Terabyte-Scale Analytics
Process terabytes of data with GizmoSQL's high-performance DuckDB engine, right on your Windows infrastructure.
90% Cost Savings
Run the same TPC-H workloads for a fraction of the cost compared to Snowflake, Databricks SQL, or BigQuery.
Full Connector Support
Works with JDBC, ODBC, Python ADBC, Power BI Connector, and all Arrow Flight SQL-compatible tools.
GizmoSQL Power BI Connector Now Available
The GizmoSQL Power BI Connector lets you connect Microsoft Power BI directly to GizmoSQL using the Arrow Flight SQL protocol. Build interactive dashboards, reports, and data visualizations powered by GizmoSQL's high-performance query engine — no intermediate data exports needed.
Key Features
DirectQuery Support
Run live queries against GizmoSQL without importing data. Your dashboards always reflect the latest data.
Hierarchical Navigation
Browse databases, schemas, and tables in the Power BI Navigator pane with intuitive drill-down.
Query Folding
Power BI pushes filters, joins, and aggregations down as SQL — including LIMIT/OFFSET, CAST, and SQL-92 expressions.
Flexible Authentication
Supports username/password, token-based, and OAuth (browser) authentication methods.
Signed Connector
The .pqx connector is code-signed for integrity verification, ensuring a secure and trusted installation.
Philip Moore Spoke at DuckDB Developer Meeting #1
GizmoData founder Philip Moore spoke at the inaugural DuckDB Developer Meeting in Amsterdam. This event focused on extension development and building sophisticated applications atop DuckDB, concentrating on the C API, extension template, and internals.
Event Schedule
| 9:30–12:00 | Extension Development Workshop | Rusty Conover (Query.Farm) |
| 15:30 | Venue opens | |
| 16:00 | Extension building in DuckDB | Sam Ansmink (DuckDB Labs) |
| 16:25 | Storage and encryption in DuckDB | Lotte Felius (DuckDB Labs) |
| 17:10 | DuckPL: A procedural language in DuckDB | Denis Hirn (University of Tübingen) |
| 17:35 | GizmoEdge: A distributed DuckDB engine for IoT GizmoData | Philip Moore (GizmoData) |
| 18:00 | Drinks and snacks | |
| 19:30 | Event ends |
Attendance is complimentary; Eventbrite registration required
Separate registration needed for the morning workshop (via Luma)
Venue is wheelchair-accessible with public transit and parking options
GizmoSQL Enterprise Edition Now Available
GizmoSQL Enterprise Edition is now available, bringing advanced features for production deployments. Built on top of the open-source Core edition, the Enterprise Edition adds session instrumentation and administrative controls for enterprise environments.
Enterprise Features
OAuth / SSO Authentication
Enterprise single sign-on support with OAuth 2.0. Integrate with your existing identity provider for seamless, secure access.
Session Instrumentation
Track server instances, client sessions, SQL statements, and query executions. Records are stored in DuckDB for analysis and auditing with views like active_sessions, session_activity, and session_stats.
KILL SESSION Command
Terminate active client sessions via SQL with KILL SESSION '<session-id>'. Requires admin role. Useful for terminating runaway queries or rogue connections.
Catalog-Level Permissions
Fine-grained access control with per-catalog read/write permissions via JWT token claims. Control which databases users can access and modify.
Includes All Core Features
GizmoSQL v1.15.1: Native Geospatial Support
GizmoSQL now includes built-in support for geospatial data! DuckDB's SPATIAL extension loads automatically at server startup with no configuration required.
Geospatial Features
GEOMETRY Type Support
Store points, lines, polygons, and more complex geometries natively in your database.
100+ Spatial Functions
Full suite of spatial functions including ST_Point, ST_Distance, ST_Contains, ST_Buffer, and many more.
GeoArrow Export
Geometry columns export with proper Arrow extension metadata for seamless interoperability.
GeoPandas Integration
Read geometry data directly into GeoPandas—no WKB conversion needed.
Example Usage
cur.execute("SELECT * FROM my_spatial_table")
gdf = gpd.GeoDataFrame.from_arrow(cur.fetch_arrow_table())
# That's it!
GizmoEdge Hits #1 on Hacker News
On October 24, 2025, GizmoEdge took the #1 spot on Hacker News. The story — "A sharded DuckDB on 63 nodes runs 1T row aggregation challenge in 5 sec" — was submitted by Tanel Poder, one of the most respected names in database performance engineering, and linked to our One Trillion Row Challenge write-up.
The result that earned the front page: GizmoEdge sharded a one-trillion-row dataset across 63 DuckDB worker nodes and ran the full aggregation challenge in about five seconds, end to end. The thread drew 224 points and a 136-comment technical discussion on distributed DuckDB, sharding strategy, and what commodity hardware can really do.
Reaching #1 on Hacker News — the front page read daily by millions of engineers — is validation from the hardest crowd there is. No ad spend, no launch coordination: just a benchmark result interesting enough that the internet's most skeptical technical audience voted it to the top.
Why It Matters
#1 on the Front Page
GizmoEdge held the top spot on Hacker News — ahead of hundreds of competing stories that day.
224 Points, 136 Comments
A deep technical discussion among distributed-systems and database engineers — not a marketing echo chamber.
Submitted by Tanel Poder
The story was posted by Tanel Poder, a world-renowned database performance expert — independent validation of the result.
1 Trillion Rows in ~5 Seconds
The underlying feat: 63 sharded DuckDB worker nodes aggregating a trillion rows in about five seconds.