Skip to content

Releases: star-ga/NikolaChess

NikolaChess v3.21.0

04 Feb 07:27

Choose a tag to compare

NikolaChess v3.21.0

100% Pure MIND chess engine - the native language for intelligent systems.

Quick Start (Linux x64)

# Download standalone binary (recommended)
curl -fsSL https://github.com/star-ga/NikolaChess/releases/download/v3.21.0/nikola-cpu -o nikola-cpu
chmod +x nikola-cpu
./nikola-cpu

Downloads

Platform Standalone Binary Status
Linux x64 nikola-cpu Ready to use
Linux ARM64 Coming soon Build in progress
macOS ARM64 Coming soon Build in progress
macOS x64 Coming soon Build in progress
Windows x64 Coming soon Build in progress

Runtime Libraries

If building from source, download the appropriate runtime library:

Platform CPU Runtime GPU Runtime
Linux x64 libmind_cpu_linux-x64.so libmind_cuda_linux-x64.so
Linux ARM64 libmind_cpu_linux-arm64.so -
macOS ARM64 libmind_cpu_macos-arm64.dylib libmind_metal_macos-arm64.dylib
macOS x64 libmind_cpu_macos-x64.dylib libmind_metal_macos-x64.dylib
Windows x64 mind_cpu_windows-x64.dll mind_cuda_windows-x64.dll

MIND Compiler

Also included: mindc v0.1.7 compiler for all platforms.

System Requirements

  • Linux: glibc 2.31+ (Ubuntu 20.04+, Debian 11+)
  • macOS: macOS 10.15+ (Catalina or later)
  • Windows: Windows 10+

UCI Protocol

Works with any UCI chess GUI (Arena, Cutechess, SCID, Lucas Chess)


Built with MIND Language v0.1.7 | Copyright (c) 2026 STARGA, Inc.

NikolaChess v3.20.0

03 Feb 12:35

Choose a tag to compare

NikolaChess v3.20.0

Supercomputer-class chess engine for the AI era.

What's New

  • Updated runtime libraries with optimized performance
  • Improved GPU memory management
  • Enhanced CPU fallback mode
  • Better Syzygy tablebase integration

Features

  • GPU-accelerated NNUE evaluation (HalfKAv2 architecture)
  • Multi-backend support: CUDA, ROCm, Metal, oneAPI, DirectX, WebGPU, CPU
  • Written 100% in MIND Language
  • Fortress-level runtime protection
  • Syzygy tablebase support: 7-man DTZ/WDL (16.7TB) / 8-man WDL (1.6PB)

Runtime Libraries (26 files)

All platform/backend combinations included:

  • Linux: x64 + ARM64 (CPU, CUDA, ROCm, oneAPI, WebGPU)
  • macOS: x64 + ARM64 (CPU, Metal, WebGPU)
  • Windows: x64 + ARM64 (CPU, CUDA, ROCm, oneAPI, DirectX, WebGPU)

Quick Install

Linux / macOS:

curl -fsSL https://nikolachess.com/install.sh | bash

Windows (PowerShell):

irm https://nikolachess.com/install.ps1 | iex

Links

NikolaChess v0.1.0

31 Jan 15:35

Choose a tag to compare

Initial release of NikolaChess - Supercomputer-class chess engine for the AI era.

Features

  • GPU-accelerated NNUE evaluation (HalfKAv2 architecture)
  • Multi-backend support: CUDA, ROCm, Metal, oneAPI, DirectX, WebGPU, CPU
  • Written 100% in MIND Language
  • Fortress-level runtime protection
  • Syzygy tablebase support: 7-man DTZ/WDL (16.7TB) / 8-man WDL (1.6PB)

Runtime Libraries (25 files)

All platform/backend combinations included:

  • Linux: x64 + ARM64 (CPU, CUDA, ROCm, oneAPI, WebGPU)
  • macOS: x64 + ARM64 (CPU, Metal, WebGPU)
  • Windows: x64 + ARM64 (CPU, CUDA, ROCm, oneAPI, DirectX, WebGPU)

Quick Install

Linux / macOS:

curl -fsSL https://nikolachess.com/install.sh | bash

Windows (PowerShell):

irm https://nikolachess.com/install.ps1 | iex

Links