Open huge CSV, Excel and JSON files right in your browser

Data files have outgrown the tools most people open them with. A nightly export from a warehouse, an ad-platform report or a payment-gateway statement routinely runs to hundreds of megabytes and several million rows — well past the 1,048,576-row ceiling of an Excel worksheet, and far past the point where a spreadsheet stays responsive. SheetPeek exists for exactly that gap: it opens the file, shows you every row, and lets you work with it.

The engine is a streaming parser paired with a virtualized grid. CSV and TSV files are read in chunks on a background worker thread, so the interface never locks up and the first rows appear almost immediately; the table itself only ever renders the few dozen rows inside your viewport, which is why scrolling through millions of records stays smooth. Excel workbooks (.xlsx and legacy .xls) and JSON arrays are decoded in memory, with a sheet picker when a workbook holds more than one sheet.

Once a file is open, the work is what you would otherwise do in a spreadsheet — only faster. Search every cell at once, or filter column by column with operators such as contains, equals, between and is empty. Sort by several columns at a time. Read per-column statistics: counts, empty cells and distinct values everywhere, min, max, mean, median and sum for numeric columns, and the ten most frequent values for everything else.

Comparing two files is a first-class feature rather than an afterthought. Point SheetPeek at yesterday’s export and today’s, choose one or more key columns, and it reports which rows were added, which were removed and which changed — down to the individual cell, with old and new values side by side. Duplicate hunting works the same way: pick the columns that define a duplicate, review every repeated record, then export a deduplicated copy. Any view you end up with leaves as clean CSV, XLSX or JSON.

None of it involves a server. There is no upload endpoint, no account, no processing queue and no stored copy of your file — the entire tool is code that runs inside your browser tab, on your own CPU and memory. That matters when the file in front of you is a payroll run, a patient list, a customer database or an unreleased financial statement: it simply never leaves the machine it is already on. Load the page once and you can disconnect from the network and keep working.