Meet Hardcharger™: fast, secure, remote DuckDB.
Hardcharger is our open-source server for DuckDB 2.0: SQL over plain HTTP with Apache Arrow results, access control built in, a web UI in the binary, and every feature included. GizmoData Cloud is the best way to run it: managed, in one click, on AWS, Google Cloud and Azure.
- A 1 TB workload derived from TPC-H in 42.4 seconds on one machine, about $0.09 of AWS on-demand compute per run
- GizmoData Cloud plans from $29/month, and the fee comes back in full as compute credit
- Users, roles, row policies, SSO and mTLS in the free build: no license keys
- Apache-2.0 open source, with drivers for Python, Go, Java and JavaScript
Launch offer: $25 of free credit, applied automatically when you sign up through the button above. No credit card needed. Rather self-host? Install Hardcharger in one line.
From the team behind GizmoSQL: 100K+ Docker pulls · In production at Summation & Randoli · Featured on Product Hunt
This workload is derived from the TPC-H Benchmark and is not comparable to published TPC-H Benchmark results, as this implementation does not comply with all requirements of the TPC-H Benchmark. TPC, TPC Benchmark and TPC-H are trademarks of the Transaction Processing Performance Council (TPC).
Everything your data team needs, fully managed
The best way to run Hardcharger: launch a managed server in one click on AWS, Google Cloud or Azure, with open lakehouse storage, a browser SQL editor, and the roles, SSO and audit trail your team needs, from one control plane.
Need it inside your own cloud account? Cloud at Customer is in production today.
A 1 TB workload, measured
Hardcharger performance, measured
Run analytical SQL over an open protocol, with DuckDB 2.0 doing the work. Here are the measured query times and raw compute costs, with the methods and caveats linked below.
- Query time
- 42.4 s
- AWS compute / run
- $0.09
- Queries × runs
- 22 × 3
Sum of the 22 per-query means, three runs per query, on one AWS Graviton5 VM. Cold first runs included. Hardcharger 2.0.0 on DuckDB 2.0.0.
Derived from TPC-H, 1 TB
Lower is better
Hardcharger uses the fresh DuckDB 2.0 run. Other systems are historical published results on different hardware and software versions; see the methodology for context.
Server costs are raw AWS on-demand compute, excluding storage and managed-service charges; not a TPC price/performance metric. AWS r9gd.24xlarge (Graviton5, 96 vCPU, 742 GiB, 3 × 1.9 TB NVMe).
This workload is derived from the TPC-H Benchmark and is not comparable to published TPC-H Benchmark results, as this implementation does not comply with all requirements of the TPC-H Benchmark. Deviations: only the 22 queries are run (no refresh functions, no power or throughput test), timed as the sum of per-query means over three runs, without an audit. TPC, TPC Benchmark and TPC-H are trademarks of the Transaction Processing Performance Council (TPC).
The same SQL. Half the bill.
Hardcharger gives you warehouse-grade analytics without the warehouse-grade invoice — on open standards you control.
Self-host or fully managed
Run Hardcharger in your own cloud — your data and your keys never leave your infrastructure — or let us operate it for you. Same engine, your call.
Benchmarks you can reproduce
Powered by DuckDB — the #2 median hot-run on ClickBench's single-node reference benchmark, ahead of Snowflake, BigQuery, Redshift & Databricks. A 1 TB workload derived from TPC-H runs in under 43 seconds for $0.09 of raw AWS on-demand compute on one machine, with Hardcharger 2.0.0.
Plugs into your stack
Speaks Chargewire (HTTP + Arrow), ADBC, JDBC, JavaScript & Python; works with dbt, DuckLake, S3 and GCS. Standard SQL, so your dashboards and queries come along.
Open by default, no lock-in
Apache-2.0, every feature included, and open table formats — no proprietary storage. Roles, row policies, SSO and audit logging come in the free build: enterprise controls, minus the enterprise invoice.
This workload is derived from the TPC-H Benchmark and is not comparable to published TPC-H Benchmark results, as this implementation does not comply with all requirements of the TPC-H Benchmark. Deviations: only the 22 queries are run (no refresh functions, no power or throughput test), timed as the sum of per-query means over three runs, without an audit. TPC, TPC Benchmark and TPC-H are trademarks of the Transaction Processing Performance Council (TPC).
Our Products
Enterprise-grade data solutions built for performance, scale, and cost efficiency
Our DuckDB 2.0 server, rebuilt from scratch as GizmoSQL’s successor. Clients connect over Chargewire, an open HTTP protocol that streams Apache Arrow, through drivers for Python, Go, Java and JavaScript or plain curl. Users, roles, grants, row policies, column masking, SSO, mTLS identity and an audit log are built in, the web UI is in the binary, and every feature is free under Apache-2.0.
Built for speed
1 TB derived from TPC-H in 42.4 s on one machine
Chargewire on the wire
SQL over plain HTTP, results as Arrow
Security in the core
Roles, row policies, SSO, mTLS
Web UI built in
Editor, schema browser, Excel upload
Using GizmoSQL? Hardcharger is its successor, and GizmoSQL 1.x is now in maintenance. GizmoSQL pages
Hardcharger 2.5.0 | Chargewire 2.0.2 | DuckDB v2.0.0
UI http://localhost:31337/ui/
Docs docs.hardcharger.gizmodata.com
✅ Hardcharger is ready
This workload is derived from the TPC-H Benchmark and is not comparable to published TPC-H Benchmark results, as this implementation does not comply with all requirements of the TPC-H Benchmark. Deviations: only the 22 queries are run (no refresh functions, no power or throughput test), timed as the sum of per-query means over three runs, without an audit. TPC, TPC Benchmark and TPC-H are trademarks of the Transaction Processing Performance Council (TPC).
GizmoData Cloud at Customer
The same platform, installed inside your own AWS, Google Cloud or Azure accounts. Managed-service experience, with your data, credentials and network never leaving your boundary.
Your account, your boundary
Control plane and every data plane inside your VPCs, IAM and KMS.
Multi-cloud from one console
EKS, GKE and AKS data planes managed by a single install.
No static keys
IRSA / Workload Identity for storage; your secret stores via external providers.
Delivered as code
Terraform in your repo, Helm from tagged, smoke-tested releases.
GizmoData sits outside your trust boundary — publishing signed releases and providing support, with no standing access to your data.
The distributed SQL engine that scales from edge devices to 1000-node clusters. Run anywhere - cloud, on-premises, or edge IoT devices.
Run Anywhere
Cloud, on-prem, or edge
1000+ Workers
Massive parallelism
Edge IoT Ready
iPhones, laptops, and more
DuckDB Core
Battle-tested engine
Distributed Architecture
1 Trillion Row Challenge
1,000 workers on Azure
What Our Customers Say
Trusted by teams building production infrastructure
We chose GizmoSQL to run our DuckLake compute engine in production. Phil has been an exceptional partner—delivering enterprise features we needed like token authentication, enhanced logging, and query timeouts, while providing fast, responsive support. Since partnering with GizmoData, we’ve gone live with one of our primary customers.
Ramachandran ("RC") Ramarathinam
Co-Founder & CTO
Summation — Decision-Grade AI Platform for Enterprise Leaders
GizmoSQL fits right into our stack — performant, flexible, and reliable. It’s a key piece of how we deliver unified observability at Randoli.
Latest News
Stay up to date with GizmoData announcements, events, and product updates
Hardcharger for Android Is on Google Play
Hardcharger for Android phones and tablets is now on Google Play, for $0.99 in the US, like the iPhone and iPad app. It runs the same hardcharger-server as Linux, macOS, Windows and iOS, with DuckDB 2.0, its web UI and a native SQL console. Share it on your Wi-Fi and Hardcharger's integrations work with it too: run a dbt project against your phone from a laptop. It's made for prototyping; for production, run Hardcharger on GizmoData Cloud.
Scheduled Jobs, Notebooks and Connections Come to GizmoData Cloud
GizmoData Cloud now runs the work around your Hardcharger clusters: scheduled jobs that copy data in from PostgreSQL, Oracle Database, Snowflake or another Hardcharger server and run SQL, Python, PySpark and dbt, JupyterLab notebooks next to your data, and saved connections to the databases you already have. Every run shows what its code steps cost.
Latest from Our Blog
Stay updated with the latest insights, benchmarks, and developments in data analytics
Schedule Pipelines, Work in Notebooks and Connect Your Databases on GizmoData Cloud
Scheduled jobs with Replication, SQL, Python, PySpark and dbt steps, JupyterLab notebooks next to your data, and saved connections to the databases you already have, with S3 access without keys. Every run shows what its code steps cost.
Bigger Ducks, Faster: Hardcharger on GizmoData Cloud vs. MotherDuck at 1 TB and 10 TB
99 queries derived from TPC-DS on a 192-vCPU Graviton5 machine: a 0.46 s geometric mean at 1 TB and 3.39 s at 10 TB, against MotherDuck Giga's published 0.69 s and 5.6 s, and ahead of every system MotherDuck cited. Per-query times, caveats and cost.
What the Heck is Hardcharger?
Shawn Gordon follows up his GizmoSQL article with its successor: an open-source DuckDB 2.0 server on Chargewire (SQL over plain HTTP, Arrow results), every former Enterprise feature included, and a local DuckDB file shared on GizmoData Cloud that Claude queries over MCP. Originally on HackerNoon.
Introducing Hardcharger™: The DuckDB 2.0 Server, Built on the Open Chargewire Protocol
Our new open-source DuckDB 2.0 server speaks Chargewire, an open protocol for SQL over HTTP with Arrow results. 22 queries derived from TPC-H over 1 TB in 42.4 s on one Graviton5 machine, about 9 cents of compute, and every feature included: RBAC, SSO, mTLS, audit and a web UI. No license keys.
Replicate Once, Query All Day: adbc-replicate Feeds AI Agents from Snowflake and Oracle
Our new open-source adbc-replicate streams Snowflake, Oracle, PostgreSQL or DuckDB schemas into GizmoSQL as Arrow over ADBC, masking inside the source and copying only what changed. Two customers, on Snowflake and Oracle, run it hourly, and their AI agents query the copy over MCP all day.
Three Clouds, One API: A 1 TB Horse Race Derived from TPC-H on GizmoData Cloud (AWS vs. Google Cloud vs. Azure)
Six 64-vCPU GizmoSQL clusters on AWS, Google Cloud and Azure raced a 1 TB workload derived from TPC-H on local NVMe and DuckLake. Graviton5 ran the fastest lap (79 s, $0.30), Google Cloud took four of six golds, and DuckDB 2.0 won every lane. 38 minutes, about $37, one API.
Introducing GizmoData Cloud: The Best Way to Run DuckDB in the Cloud
GizmoData Cloud is generally available: fully managed GizmoSQL clusters, DuckLake catalogs and federated queries on AWS, Google Cloud and Azure. Sign up free. Start free with $25 of launch credit, applied automatically, no credit card.
The Connection That Never Left: Keeping GizmoSQL Lean With Session Limits and Idle Timeouts
Nobody ran a query since Friday, yet Monday morning the server is at 80% memory: the queries left, the connections didn't. Guest author Emmanuel Sibanda on --max-sessions and --session-idle-timeout, and why only real user SQL resets the idle clock.
GizmoSQL + DuckLake: Elastic Compute Over Shared State
Scaling nodes that share mutable data is the hard part of analytics. With DuckLake holding the transactional catalog and Parquet in object storage, a GizmoSQL cluster grows from one node to five and back with nothing to copy, sync, or reconcile.
Seagull: How to Use GizmoSQL to Build Scalable Web GIS Applications
Aalborg University's Seagull platform serves 18.7 billion AIS records as live map layers, every tile computed on request from DuckDB. GizmoSQL made single-writer DuckDB a shared database for Airflow, MapLibre Martin, and the web app.
1 TB in 80 Seconds, Derived from TPC-H: GizmoSQL on AWS Graviton5 (vs. Graviton4 and Azure Cobalt 100)
A 1 TB workload derived from TPC-H (22 queries) on a brand-new Graviton5 r9gd.24xlarge: 80.5 seconds, 10% faster than Graviton4 and 25% faster than Azure Cobalt 100 in same-day, same-DuckDB runs, still $0.17 of compute. Methodology and every per-query timing.
Introducing the GizmoSQL MCP Server: Talk to Your DuckDB Data from Claude
An MCP server and Claude Desktop extension for GizmoSQL — Claude Desktop, Claude Code, and any MCP client can explore your schema and run SQL, read-only by default with server-enforced row caps and timeouts.
Meet the GizmoData ADBC Driver Suite: Move Data Between Oracle, Db2, and GizmoSQL
Pure-Go, pip-installable ADBC drivers for Oracle (no Instant Client), IBM Db2 (no CLI/JCC), and DuckDB's Quack — alongside our flagship GizmoSQL driver. Arrow-native, so moving a table between them is a few lines of Python or one SQL statement, in both directions.
The Arrow Scan Rewrite: How GizmoSQL v1.37.0 Made Bulk Ingest 7x Faster
GizmoSQL v1.37.0 stopped hand-converting every Arrow cell and started streaming batches through DuckDB's own arrow_scan instead — 7.2x faster bulk loads, exact decimals, and geometry columns that finally land as GEOMETRY.
GizmoSQL ADBC Driver 2.0: A Go-Powered Core That Brings GizmoSQL to Every ADBC Language
The GizmoSQL ADBC driver has been rebuilt in Go with a thin, drop-in-compatible Python wrapper — bringing OAuth/SSO, automatic DDL/DML detection, and more to every ADBC language.
Welcoming Quack: DuckDB's New Client-Server Protocol — and Where GizmoSQL Goes From Here
DuckDB Labs announced Quack, a native client-server protocol for DuckDB. A thank-you to the team — and a look at how GizmoSQL complements the growing client-server DuckDB ecosystem.
Freeing the Individuals from the Matrix: How GizmoSQL Makes Your Database AI-Agent Ready
A Matrix-themed deep dive into building an AI-agent-ready architecture with GizmoSQL, DuckDB, and Claude—simulating one million people and monitoring anomalies with an autonomous AI agent.
ADBC Flight SQL 1.10.0: Bulk Ingestion for GizmoSQL
The new ADBC Flight SQL Python driver 1.10.0 brings bulk ingestion to Arrow Flight SQL servers like GizmoSQL. Load millions of rows directly from Python or DuckDB with automatic schema inference.
What the Heck is GizmoSQL?
A deep dive into GizmoSQL—a small server that runs DuckDB with the Arrow Flight SQL protocol wrapped around it so you can run DuckDB remotely.
Introducing GizmoSQL Operator
Announcing the GizmoSQL Kubernetes Operator—a tool to reduce operational costs of running large-scale SQL queries.
GizmoEdge Takes on the 1 Trillion Row Challenge
What happens when you give a distributed SQL engine a trillion-row dataset? GizmoEdge crushed the Coiled challenge with a 1,000-worker cluster on Azure.
GizmoSQL One-Trillion Row Challenge Results
GizmoSQL completed the 1 trillion row challenge! Learn how we processed massive datasets using DuckDB and Apache Arrow Flight SQL.
Articles
Insights and technical knowledge from our team
Powering Data APIs with Arrow Flight RPC and Ibis
October 25, 2024 • by Philip Moore
In this article, we'll demonstrate how you can use a Apache Arrow Flight RPC API server powered by Ibis to create fast, efficient data APIs.
Featured Videos
See our products in action
Run DuckDB on Your iPhone! Exploring GizmoSQL's Lightning-Fast Mobile Data Warehouse
Independent coverage of the GizmoSQL iOS app — running DuckDB as a lightning-fast mobile data warehouse, right on your iPhone.
Philip Moore on the A/I Prosperity Podcast
GizmoData founder Philip Moore joins Torben Andersen to talk GizmoSQL, running DuckDB as an enterprise Arrow Flight SQL server, and cutting cloud warehouse costs.
TPC-H Benchmarks on GizmoSQL iOS, Fast 🚀
Running TPC-H benchmarks on GizmoSQL iOS — see the blazing-fast performance.
Introducing the New GizmoSQL iOS App
See the brand-new GizmoSQL iOS app, now available on the Apple App Store.
GizmoEdge at the DuckDB Developer Meeting
Presenting GizmoEdge at the inaugural DuckDB Developer Meeting in Amsterdam
Introducing the New Windows Release of GizmoSQL
We demonstrate how easy it is to run GizmoSQL - now available for Windows.
GizmoSQL in action
Run 10 TPC-H SF100GB benchmarks, at the same time!
GizmoEdge
The world's first IoT-powered distributed SQL engine
Contact Us
We know that our clients have unique needs. Get in touch, and we will get back to you soon.
Subscribe to Our Newsletter
Stay updated with the latest news, features, and insights from GizmoData.
Free, and you can unsubscribe at any time. Pressing Subscribe loads the sign-up form from Substack, which sends the newsletter. Read past issues