Best Parquet viewers in 2026: free and paid options compared

Pick by where the viewer runs, what it costs, its main catch, and whether you can query the file with SQL.

rows.page

rows.page opens .parquet and .pq files in your browser with DuckDB-WASM, without the 1.5 GB cap it applies to CSV and JSON, and nothing is uploaded unless you press Share. You get per-column stats, DuckDB SQL on a table called data, and exports to CSV, JSON, JSONL, Markdown or Parquet. The catch: it is read-only, and SQL results stop at 1,000,000 rows. Opening files is free with no signup; hosted links run on Free ($0), Pro ($12 per month) or Max ($39 per month).

SELECT country, count(*), avg(price)
FROM data
WHERE price > 20
GROUP BY ALL

parquet-viewer.com

parquet-viewer.com reads Parquet files locally in your browser with WebAssembly and shows the schema, statistics and metadata. The free tier lets you browse, filter, sort and export, while the full SQL editor and uncapped exports are Pro features. Pro costs $59 for one year with no automatic renewal, or $4.99 for a day pass.

Hugging Face dataset viewer

The Hugging Face dataset viewer shows datasets hosted on the Hub in pages of 100 rows, and its SQL Console runs DuckDB queries in your browser. Filter, search and statistics cover only the first 5 GB of a dataset. It is free, but it works on datasets on the Hub, not on a file on your disk.

Tad

Tad is a desktop app for macOS, Windows and Linux with DuckDB inside, built around a pivot table with histograms and filtered exports. There is no SQL editor: you pivot, filter and sort in the UI. It is free (MIT), and the latest release, 0.14.0, came out in June 2024.

ParquetViewer

ParquetViewer is a native Windows app for opening Parquet files and reading their metadata. You can filter rows with simple SQL WHERE conditions, but it does not run full queries, and there is no macOS or Linux version. It is free (GPL-3.0).

DuckDB UI

Run duckdb -ui (DuckDB 1.2.1 or newer, since March 2025) and a local SQL notebook opens at localhost:4213. It is built by MotherDuck, runs your queries on your machine, and its frontend is not open source. It needs the DuckDB CLI, does not support Intel Macs or Windows on ARM, and is free.

DBeaver

DBeaver can open Parquet files next to your databases and query them with SQL. The Parquet driver is read-only and ships only in the paid Lite, Enterprise and Ultimate editions, not in the free Community edition.

Parquet Viewer by dvirtz

This extension shows a .parquet file as JSON or CSV text in an editor tab. VS Code lets extensions work only on files under 50 MB, so bigger files are truncated, and there is no SQL. It is free (MIT).

Data Wrangler

Microsoft's Data Wrangler opens .parquet files in VS Code as a grid, and each sort or filter you apply generates pandas code. It needs Python 3.8 or newer (it installs pandas for you), and it has no SQL. It is free.

parquet-cli

parquet-cli is the Java tool from the Apache Parquet project that prints a file's schema, metadata and first records. Use it instead of the older parquet-tools, which was removed from apache/parquet-java. It is free (Apache-2.0).

pqrs

pqrs is a Rust binary that prints a Parquet file's schema, row count and records as text, JSON or CSV, with no Java runtime needed. It is free (Apache-2.0 or MIT).

• Quick look with SQL and no install: rows.page, free with no signup and no 1.5 GB cap on Parquet.
• A dataset already on the Hugging Face Hub: its dataset viewer and SQL Console.
• A desktop app: Tad on any of the 3 platforms, ParquetViewer on Windows, or DBeaver if you already pay for it.
• A local SQL notebook: duckdb -ui with DuckDB 1.2.1 or newer.
• Inside VS Code: the dvirtz Parquet Viewer for files under 50 MB, or Data Wrangler for pandas work.
• A server over SSH: pqrs or parquet-cli.