whereismyai
L4h
L4h: Type-1 hypervisor that runs Linux virtual machines on the seL4 microkernel.
L4H is a Type-1 hypervisor built on the seL4 microkernel that runs the Linux kernel as usermode threads without hardware virtualization. It targets 2020+ Intel and AMD CPUs on bare metal and KVM-based VMs. The system uses a single EFI image containing seL4, a guest Linux kernel, and system component processes like root, vmmgr, vmm, trustd, blkd, and netd. It protects VM integrity so that dom0 lacks full control over guests. It supports TPM-backed remote attestation and IOMMU protection for block and network drivers.
L4h at a glance
- Best for: Software engineer
- Runs on: self-hosted, open source
- Pricing: Free
- Released: October 2026
- Category: AI coding tool
- Website: github.com/losfair/l4h
Key features
- x86-64 multi-core guest support
- TPM-backed remote attestation
- Memory hotplug
- In-kernel AArch64 emulation
- Virtio protocol for emulated devices
- IOMMU protection for block and network drivers
Use it for
- Running isolated Linux virtual machines
- Managing VM lifecycle with fine-grained control
- Providing secure block and network devices to VMs
- Generating TPM-backed VM attestation
Alternatives to L4h
- llm-d — Deploys production-grade LLM serving on existing hardware clusters.
- vLLM Ascend — It enables vLLM functionality on Huawei Ascend NPU hardware.
- Sandkiln — Runs untrusted code in hardware-isolated Firecracker microVMs.
- CubeSandbox — Provides sub-60ms cold starts and hardware-level microVM kernel isolation.
- llmrun — Check which open models run on your specific GPU or Mac instantly.
- LangChain4j — Integrates LLMs and vector stores directly into Java applications.
- Pebrel — It brings local shells, remote hosts, and AI command-line tools into one workspace.
- Hlock — Pins entire AI agent setups into a portable file to share and run anywhere.
- Tunr — Deploys agent-built software and shares local dev servers instantly via a single command.
- RunThisLLM — Provides exact hardware and VRAM requirements for over 295 local AI models.
- ssh zydra.sh — Run AI agents overnight in a sandboxed virtual machine with built-in Claude Code and Codex.
- Daytona — It provisions sandboxes in under 90 milliseconds with full security isolation.
All L4h alternatives