How to open a Parquet file

Method 1

Open in your web browser (recommended for one-off inspection)

The fastest option, no install required. Open the Parqui online viewer and drag your .parquet file in. The viewer parses the file entirely in your browser— the bytes never leave your machine — so it's safe for confidential data.

You get virtual scrolling, column filtering, multi-column sort, full-text search, and one-click export to CSV or JSON. Files of several hundred megabytes work smoothly; for multi-GB files, prefer the desktop app.

When to use: quick inspection, sharing a screenshot, confirming schema or row counts, exporting to CSV/JSON.

Method 2

Use the Parqui desktop app (best for large files and offline use)

Download Parqui for Windows, macOS, or Linux. Double-clicking any .parquet file in your file manager opens it directly. The desktop app handles multi-gigabyte files comfortably and works offline. Built with Tauri, so the binary is small and fast.

When to use: daily work with Parquet files, large files (several GB+), no internet access, native OS file association.

Method 3

Open in Python with pandas or pyarrow

For data analysis, the standard tool is pandas:

import pandas as pd

df = pd.read_parquet("data.parquet")
print(df.head())
print(df.dtypes)

For lower-level control, use pyarrow directly — useful when you want to inspect the schema or only read specific columns:

import pyarrow.parquet as pq

table = pq.read_table("data.parquet", columns=["user_id", "amount"])
print(table.schema)
print(table.num_rows)

When to use:you're already in a Jupyter notebook, ML pipeline, or Python script.

Method 4

Query with SQL via DuckDB (no setup, no Python)

DuckDB queries .parquet files directly — no schema definition, no importing, no database setup:

duckdb -c "SELECT category, COUNT(*) AS n
           FROM 'data.parquet'
           GROUP BY 1 ORDER BY n DESC LIMIT 10"

Works on local files, S3 paths, and even directories of Parquet files. The single binary is ~30 MB and there's nothing to install or configure beyond that.

When to use:you want to ask SQL questions, you have many .parquet files at once, or you're joining Parquet with other data sources.

Method 5

Embed a viewer in your own app

If you're building a data tool, internal admin panel, ML annotation UI, or HuggingFace-style dataset preview, drop a Parqui component into your app:

npm install @parqui/react

import { Parqui } from "@parqui/react";

export default function App() {
  return <Parqui source="https://example.com/data.parquet" />;
}

Same component is available for Vue and Angular. You get virtual scrolling, filtering, sorting, and CSV/JSON export out of the box. Free for personal and open-source projects; commercial licensing starts at €49.

When to use:you're building a product or internal tool that needs to display Parquet data to end-users.