What is steam-catalog-data?

This repository is a dataset, not software you install and run. It holds open data about the top 10,000 games on Steam, the popular PC gaming store, gathered for a study called The State of Steam 2026. Every piece of information comes from public storefront listings or from SteamSpy, a third party service that estimates how many people own each game, so no personal data about individual players is included. The data is provided in a few different formats depending on how you want to use it. A single JSON file holds all the summarized statistics behind the study, things like pricing trends by year, how games are distributed across review score categories, popular genres, and how ownership is concentrated among the most popular titles. A sample file shows 200 individual game records in full detail, giving a sense of what the complete underlying dataset looks like, including fields like price, developer, publisher, review counts, playtime, supported platforms, and release date. Several smaller CSV files break out specific slices of the same information, useful for quickly loading into a spreadsheet program or a data analysis tool. The README highlights several findings from the study itself. Ownership is very concentrated, with the most popular 1 percent of games accounting for roughly a third of all estimated game ownership across the platform. New game prices have been rising while free to play releases have become less common in recent years. Support for Mac and Linux operating systems has declined sharply over the past decade, even though Steam's own Steam Deck handheld device runs Linux. Most top games also receive positive reviews, and independently made games make up a large share of the top titles. The dataset covers the top 10,000 games by estimated ownership, not Steam's full catalog of tens of thousands of titles, and ownership numbers are estimates rather than exact sales figures. The data was captured as a snapshot in July 2026. This project is licensed under Creative Commons Attribution 4.0, meaning anyone can use the data freely as long as they credit the original source.