The Lichess Open Database is one of the largest freely available chess game datasets in the world, containing hundreds of millions of games played on Lichess going back to 2013. All data is released into the public domain with no restrictions, making it an invaluable resource for researchers, developers, and analysts.

Games are organized into monthly archives and available in PGN format (standard-rated and variant games separate), as well as a custom compressed format (Zstandard/zstd) for efficient bulk downloads. Each game includes metadata such as player names, ratings, time controls, opening ECO codes, and engine evaluations where applicable.

This database has powered countless chess research papers, machine learning projects, and third-party tools. If you want to train a model, run statistical analysis, or build an opening explorer from real games, the Lichess database is the starting point of choice.