[!CAUTION] This dataset is still a work in progress and some breaking changes might occur.

  Lichess Rated Standard Chess Games Dataset





  Dataset Description

6,771,826,271 standard rated games, played on lichess.org, updated monthly from the database dumps. This version of the data is meant for data analysis. If you need PGN files you can find those here. That said, once you have a subset of interest, it is trivial to convert it back to PGN as shown in the Dataset Usage… See the full description on the dataset page: https://huggingface.co/datasets/Lichess/standard-chess-games.

1) Data Introduction • The Chess Game Dataset (Lichess) includes details of more than 20,000 chess matches collected on Lichess.org , as well as players, openings, and match results.

2) Data Utilization (1) Chess Game Dataset (Lichess) has characteristics that: • This dataset provides a variety of variables related to chess matches, including match ID, match start and end times, turn count, winner, player rating, opening code and name, and full sequence. (2) Chess Game Dataset (Lichess) can be used to: • Openings Win Rate Analysis: Analysis of the frequency of use and win rate of each opening can be used to study effective chess strategies. • Player skill prediction: Based on player ratings, match results, and sequential data, it can be used to predict wins and losses and analyze performance improvement factors.

Dataset Card for Lichess Puzzles

  Dataset Description

5,423,662 chess puzzles, rated and tagged. See them in action on Lichess. This dataset is updated monthly, and was last updated on October 7th, 2025.

  Dataset Creation

Generating the initial dataset chess puzzles took more than 50 years of CPU time. We went through 300,000,000 analyzed games from the Lichess database, and re-analyzed interesting positions with Stockfish 12/13/14/15 NNUE at 40 meganodes. The… See the full description on the dataset page: https://huggingface.co/datasets/Lichess/chess-puzzles.

Dataset Card for the Lichess Evaluations dataset

  Dataset Description

302,517,109 chess positions evaluated with Stockfish at various depths and node count. Produced by, and for, the Lichess analysis board, running various flavours of Stockfish within user browsers. This version of the dataset is a de-normalized version of the original dataset and contains 752,452,094 rows. This dataset is updated monthly, and was last updated on Thursday 16th, 2025.

  Dataset… See the full description on the dataset page: https://huggingface.co/datasets/Lichess/chess-position-evaluations.

A collection of 25,940 chess games played by the user 'pawnster101' on the online platform Lichess.org. The dataset includes game data and player interactions in a clean interface without ads or registration requirements.

General Info

This is a set of just over 20,000 games collected from a selection of users on the site Lichess.org, and how to collect more. I will also upload more games in the future as I collect them. This set contains the:

• Game ID;
• Rated (T/F);
• Start Time;
• End Time;
• Number of Turns;
• Game Status;
• Winner;
• Time Increment;
• White Player ID;
• White Player Rating;
• Black Player ID;
• Black Player Rating;
• All Moves in Standard Chess Notation;
• Opening Eco (Standardised Code for any given opening)
• Opening Name;
• Opening Ply (Number of moves in the opening phase) For each of these separate games from Lichess. I collected this data using the Lichess API, which enables collcollection of any given users' game history. The difficult part was collecting usernames to use, however the API also enables dumpdumping of all users in a Lichess team. There are several teams on Lichess with over 1,500 players, so this proved an effective way to get users to collect games.

Possible Uses

Lots of information is contained within a single chess game, let alone a full dataset of multiple games. It is primarily a game of patterns, and data science is all about detecting patterns in data, which is why chess has been one of the most invested in areas of AI in the past. This dataset collects all of the information available from 20,000 games and presents it in a format that is easy to process for analysis of, for example, what allows a player to win as black or white, how much meta (out-of-game) factors affect a game, the relationship between openings and victory for black and white and more.

This is a collection of ~1.5M chess puzzles from the Lichess database of ~3.9M puzzles (as of 2024-05-09). The set of puzzles from "Can You Learn an Algorithm? Generalizing from Easy to Hard Problems with Recurrent Networks" is included, with the exception of 26,079 puzzles that are no longer in the Lichess database (on the assumption that they might have been removed for a good reason). For each puzzle, ctx is a SAN transcript (with every half-move numbered) of an actual Lichess game, up to… See the full description on the dataset page: https://huggingface.co/datasets/EleutherAI/lichess-puzzles.

Dataset Card for Lichess Puzzles

  Dataset Description

3546 chess openings with their Encyclopaedia of Chess Openings (ECO) classification.

  Dataset Creation

Creating this dataset is detailed in its original GitHub repository. Updates to the original repo will also be reflected in this version. Dataset last updated on October 7th, 2025.

  Dataset Usage

Using the datasets library: from datasets import load_dataset

dset =… See the full description on the dataset page: https://huggingface.co/datasets/Lichess/chess-openings.

Traffic analytics, rankings, and competitive metrics for lichess.org as of August 2025

This dataset is a comprehensive collection of chess positions sourced from Lichess, one of the most popular online chess platforms. This dataset offers a valuable resource for researchers, enthusiasts, and developers interested in exploring and analyzing chess games.

Key Features

• Chess Positions: The dataset comprises 983,014 unique chess positions obtained from Lichess. Each position represents a snapshot of a game at a specific moment and includes the placement of chess pieces on the board.
• Stockfish 15.1 Scores: Every chess position in the dataset is accompanied by a score assigned by Stockfish 15.1, a powerful open-source chess engine renowned for its exceptional playing strength. The scores provide an assessment of the position's relative strength, with positive values indicating an advantage for white and negative values indicating an advantage for black.
• Additional Information: Alongside the chess positions and Stockfish scores, the dataset includes supplementary information associated with each position. This information includes usernames, elo ratings of the players, and the final results of the corresponding games, among other relevant details. This data enriches the dataset and enables various analyses, such as studying the performance of specific players or exploring the relationship between player ratings and position evaluations.

Potential Use Cases

• Chess AI Development: Researchers and developers interested in developing or improving chess-playing algorithms can leverage this dataset to train machine learning models. The combination of chess positions and associated Stockfish scores offers a valuable resource for training AI models to make informed decisions and evaluate the strength of chess positions.
• Chess Strategy Analysis: Chess enthusiasts can utilize this dataset to study and analyze various chess positions and the corresponding evaluations by Stockfish 15.1. By examining positions where the engine assigns significantly higher or lower scores, users can gain insights into strategic moves, patterns, and potential winning strategies.
• Player Performance Evaluation: The additional information provided in the dataset, such as usernames and elo ratings, allows for an analysis of player performance. Researchers can explore how different players' strategies and decision-making impact the Stockfish evaluations, shedding light on various playing styles and approaches.

Note: While some positions may appear multiple times in the dataset, they represent distinct moments within different games. The dataset encourages a comprehensive understanding of chess dynamics by offering a diverse range of positions and evaluations.

Dataset

This dataset was created by Seqaeon

Contents

Context

It all started with Covid-19 pandemic. The world got home struck. I started watching videos of a lot of content creators in Youtube. I started watching a stand-up comedian turned streamer's videos and he introduced chess and lo-and-behold, my fond for chess started!!

Content

The data in this dataset are all stats of my chess matches that I've played either in the morning or in the evening, with my friends or randoms. All the data is mine but I have collected it from Lichess.org, an online chess playing and learning webapp. All the data is manually written by my ownself!

Acknowledgements

Thanks to Lichess.org for having come up with such a thought that I was able to access this wonderful place.

Inspiration

I'm facing issues as of now crossing 1000 rating in 5 min + 0 increment Blitz. So, what can I do, what opening's should I go for, which pieces I'm good with and etcetera.

🧠 About This Dataset

This dataset showcases synthetic chess commentary generated from real Lichess games, designed to empower research in LLMs, sports analytics, and game understanding.

🚀 Dataset Creation Pipeline

• Game Source: Real match data fetched from Lichess.org
• Analysis Engine: Each game was processed using Stockfish, extracting turn-by-turn evaluations and chess insights
• Commentary Generator: These insights were passed to Gemini API, producing commentary in Alpaca-style JSON format, structured for LLM compatibility

🧩 Dataset Contents

Each JSON object contains: - Unique match ID - Player moves and board states - Stockfish's turn-level analysis - Synthetic commentary in alpaca format with each move

📈 Intended Use Cases

This dataset supports: - Fine-tuning LLMs for dynamic game commentary - Evaluating generative models in sports settings - Building intelligent chess dashboards or companion apps

⚙️ Format & Licensing

• Format: Single JSON file with game annotated matches
• License: Apache 2.0—open for reuse and adaptation

Condensed Lichess Database

This dataset is a condensed version of the Lichess database. It only includes games for which Stockfish evaluations were available. Currently, the dataset contains the entire year 2023, which consists of >100M games and >2B positions. Games are stored in a format that is much faster to process than the original PGN data.

Requirements: pip install zstandard python-chess datasets

  Quick Guide

In the following, I explain the data format… See the full description on the dataset page: https://huggingface.co/datasets/mauricett/lichess_sf.

Dataset

This dataset was created by Seqaeon

Contents

[!CAUTION] This dataset is still a work in progress and some breaking changes might occur. In the meantime, please use https://database.lichess.org/#variant_games

Context

This dataset is contains more than 7 million chess matches between players above 2200+ rating on lichess excluding bullet games

Content

This contains folder which is monthly rapid games played on lichess starting from june-2020 all upto june-2021 in pgn format

Acknowledgements

We wouldn't be here without the help of others. If you owe any attributions or thanks, include them here along with any citations of past research.

Inspiration

It was collected from li chess to make a chess engine capable of making decent moves and then self learn to improve

[!CAUTION] This dataset is still a work in progress and some breaking changes might occur.

  Note

The FEN column has 961 unique values instead of the expected 960, because some rematches were recorded with invalid castling rights in their starting FEN in November 2023.

Lichess Broadcasts

This is a dataset of chess games from chess tournaments tracked using Lichess Broadcasts. Lichess Broadcasts show live games as they unfold with new moves arriving in real time. They are built to connect to the live-updating PGN file produced by DGT boards but can work with other sources as well. Broadcasts are organized in "tournaments" and "rounds."

[!CAUTION] This dataset is still a work in progress and some breaking changes might occur. In the meantime, please use https://database.lichess.org/#variant_games