CSSLab/maia-chess

Health v2 · maintenance only

Adoption not part of the score

1228 stars · 151 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Maia is a collection of human-like neural network chess engines trained on millions of human games, targeting skill levels from ELO 1100 to 1900. The repo provides model weights (Leela Chess format) plus code to train new models and reproduce results, requiring lc0 as the inference engine.

Use cases

• play chess against an AI that plays like a human at a given skill level
• download human-like chess engine weights for lc0
• train a neural network to predict human chess moves
• research human behavior and mistakes in chess across skill levels
• run a chess bot on Lichess that mimics human play
• analyze games from a human-like perspective instead of engine-optimal moves

When to choose

• you want a chess opponent that makes human-like moves and mistakes at a target ELO
• you need pretrained chess neural network weights for lc0
• you are studying human decision-making or skill modeling in chess
• you want to build a Lichess bot that feels like a real person

When to avoid

• you need the strongest possible chess engine for analysis or competition
• you want a full-featured chess GUI or framework - Maia only provides model weights and needs lc0
• you need a self-contained engine binary that runs without external dependencies

Facets

library · maturity active

machine-learning deep-learning game-engine machine-learning artificial-intelligence python cross-platform chess chess-engine neural-network human-like-ai lc0 elo-modeling computational-social-science game-development

Member repositories

For agents

markdown · JSON · MCP: product_card(name="CSSLab/maia-chess")