# lightvector/KataGo

GTP engine and self-play learning in Go

Repository: https://github.com/lightvector/KataGo
Canonical: https://ross.abutalabs.com/products/katago
Homepage: https://katagotraining.org/
Language: C++
License: NOASSERTION
License Family: other
Last push: 2026-08-25T16:33:19+00:00

## Health v2 (maintenance only)
Score: 99/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 99, longevity 100
- inputs: {"age_days": 2745, "days_push": 8, "days_rel": 9, "gap_med": 20.5, "n_releases_24m": 11}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5036, forks 752 (observed 2026-08-28T04:09:08.700894+00:00)

## What it is
KataGo is an open-source Go (baduk) engine trained via AlphaZero-like self-play, one of the strongest Go bots available. It runs as a GTP engine and analysis engine, with support for GPU backends like CUDA, TensorRT, OpenCL, ROCm, ONNX, and Eigen.

## Use cases
- play Go against a strong AI bot
- analyze my Go game records and get win-rate evaluations
- train a Go AI from scratch with self-play
- run a GTP engine with a Go GUI like Sabaki or Lizzie
- contribute GPU compute to a distributed Go training run
- research faster AlphaZero-style self-play training methods

## When to choose
- you need a top-strength open-source Go engine for play or analysis
- you want to study or extend self-play neural net training for board games
- you need handicap play, score prediction, and multiple board sizes and rulesets in one engine

## When to avoid
- you need an engine for chess, shogi, or other games (use Stockfish or Leela variants)
- you have no GPU and need maximum playing strength
- you want a graphical interface out of the box - KataGo is a backend engine

## Facets
- artifact type: application
- maturity: active
- function: machine-learning, deep-learning, llm-training, simulation, game-engine
- domain: artificial-intelligence, machine-learning, deep-learning, gpu-computing
- platform: cpp, windows, cli
- tags: go-game, gtp-engine, self-play-training, alphazero, board-games, neural-network, mcts, distributed-training, game-development, linux, macos, gpu

## Member repositories
- lightvector/KataGo (main) score 99

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:08.700894+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:17:25.988816+00:00, confidence not recorded.
  - readme: https://github.com/lightvector/KataGo (fetched 2026-08-28T04:09:08.700894+00:00, sha 7d6745eabe50)
  - homepage: https://katagotraining.org/ (fetched 2026-08-29T08:57:59.013939+00:00, sha af1dcd37d7c0)
- Data as of 2026-08-30T08:39:29.467469+00:00.