MerlionOS

Born for AI. Built by AI. — A hobby OS kernel written in Rust for 4 CPU architectures.

Why MerlionOS?

Written in Rust

Memory-safe kernel with zero undefined behavior. No C, no compromises.

AI-Native Architecture

NL shell, neural net inference, ML training, vector store, AI workflows, and self-evolution built into the kernel.

330 Modules, 121K Lines

Comprehensive kernel: memory, multitasking, SMP, filesystems, TCP/IP, servers, GPU, audio, GUI, WiFi, Bluetooth, and more.

4 CPU Architectures

Runs on x86_64, aarch64 (Raspberry Pi), RISC-V, and LoongArch — with shared kernel core and per-arch HAL.

450+ Shell Commands

Rich interactive environment: vim, bash/zsh, ss, nc, ip, systemctl, traceroute, sysctl, and full Linux-style tooling.

At a Glance

Preemptive multitasking with per-process page tables, signals, and process groups
SMP support with APIC, HPET timers
Filesystems: VFS, FAT16, ext2, ext4, tmpfs, procfs, sysfs
Storage: virtio-blk, AHCI, NVMe, GPT, NTFS read-only, software RAID 0/1/5
Networking: TCP (Reno/Cubic/BBR), UDP, IPv4+IPv6, DHCP, DNS, TLS, WebSocket, MQTT, QUIC/HTTP/3, gRPC, OSPF/BGP, iptables/NAT, eBPF, DPDK
Servers: HTTP, SSH, DNS, MQTT broker, HTTPS reverse proxy
Security: Capabilities, seccomp, ACLs, AES-128, RSA, X.509
Display: Window compositor, desktop environment, framebuffer terminal
Input: USB mouse, keyboard layouts, touchpad
Hardware: PCI, USB (xHCI), Bluetooth, e1000e NIC, ACPI power management, S3 sleep/wake
Languages: Forth, Lisp, WASM/WASI runtime, shell scripting
Multi-architecture: x86_64, aarch64 (ARM), riscv64 (RISC-V), loongarch64 (LoongArch)
UEFI boot via Limine — runs on real hardware
87 releases from v1 through v87

Projects

merlion-kernel

General-purpose hobby OS kernel. 330 modules, 4 architectures, 450+ shell commands. Get Started | GitHub

merlion-infer

Bare-metal OS for LLM inference. Zero overhead, OpenAI-compatible API, AMD GPU support. Learn More | GitHub

Quick Start

merlion-kernel (General Purpose)

git clone https://github.com/MerlionOS/merlion-kernel.git
cd merlion-kernel
make run

merlion-infer (Inference Server)

git clone https://github.com/MerlionOS/merlion-infer.git
cd merlion-infer
./tools/download_model.sh  # Download SmolLM-135M
make run-full              # Boot with disk + network
# merlion> ai-serve 8080   # Start OpenAI API