<\!DOCTYPE html> Zero — The OS the AI era demands <\!-- Inter + IBM Plex Mono -->
Ring-0 Native AI Operating System

The Windows
of AI.

Zero is the first operating system where AI is the kernel — not an application running on top of Linux. Ring-0 execution. Bare metal. No syscall overhead. 149.3 tok/s on AMD EPYC.

Get Zero — Free How it works →
149.3
tok / sec
CPU-only on AMD EPYC 9354P
10×
faster
than typical llama.cpp desktop
0
Linux
bare metal Ring-0 execution
Zero OS — Ring-0 AI Kernel Architecture
Ring-0 execution
The Problem

For 50 years, operating systems were built for humans typing at terminals.

The architecture we inherited — kernel/user space splits, syscall overhead, process isolation — was designed for a different era. Running AI on Linux is like racing a Formula 1 car on a dirt road. The car is fast. The road is the bottleneck. Zero is the road built for the car.

Legacy Stack
Application Layer
Host OS (Linux / Windows)
Virtualization / Hypervisor
Hardware
4 context switches per inference call
Zero OS
AI Kernel (Ring-0)
Hardware Abstraction Layer
Bare Metal
Zero syscall overhead
Architecture

No user-space. No kernel/user split.
No process boundaries. No syscalls.

<\!-- Light-theme architecture SVG --> <\!-- Background --> <\!-- Ring-3 box --> RING-3 User Space Sandboxed Agents AgentLang Runtime App Binaries (AOT) Type-safe via Validator + Capability system <\!-- Arrow Ring3 → Ring0 --> <\!-- Ring-0 box (main kernel) --> RING-0 AI Kernel — The OS Core <\!-- Boot-LLM sub-box --> Boot-LLM Qwen3-1.7B Q4_K_M 149.3 tok/s (CPU-only) x86_64 + ARM64 deterministic <\!-- Validator sub-box --> Validator Type-safe code execution 270+ test cases passing No runtime JIT overhead <\!-- Bottom row kernel sub-components --> Arena Allocators O(1) bump-pointer Zero-Copy IPC Capability-typed handles AOT Codegen Native x86_64 <\!-- HAL strip --> Hardware Abstraction Layer — x86_64 · ARM64 · HVF <\!-- HAL down-arrow -->
Safety enforced by the Validator and type system — not page tables or hardware rings
Single Ring-0 address space — no kernel/user split, no context switch cost
Unikernel architecture — one address space, one purpose: AI inference at hardware speed
Product Family

The OS that AI runs on.
Free and paid editions.

Zero Basic and Zero Hardcore are 100% free, AGPLv3, shipped complete — not a trial. Paid tiers add production tooling, SMP support, and enterprise compliance. Built by Ring0.

<\!-- Product cards -->
Free
Zero Basic
Everyday local AI OS. Runs on consumer PCs, laptops, and EPYC hardware. Perfect for individuals and developers.
AGPLv3 · Open Source
Free
Zero Hardcore
Bare kernel, no abstraction layers. For builders, hackers, and OS researchers who want the raw Ring-0 primitives.
AGPLv3 · Open Source
Free
Zero LLM Standard
Built-in inference for every Zero user. Qwen3 Q4_K_M at 149.3 tok/s on EPYC. Included with every edition.
Always included
Paid
Zero LLM Premium
Rent frontier model access via Zero: Anthropic, OpenAI, Gemini — routed natively through the OS without a host.
$/month rental
Paid
Zero Server
EPYC and Xeon optimized. Native NICs, full SMP, production tooling, priority support and SLA. For production deployments.
Commercial license
Paid
Ring0 Cloud
Server rental with Zero pre-installed. Bring your own model. Managed infrastructure, zero configuration.
$/month hosting
Free / Paid
Ring0 Benchmark
Self-hosted benchmark tool: free. Cloud-ranked leaderboard with verified results: paid. Prove your hardware.
Free self-hosted · Paid cloud
Enterprise
Zero Certified
Compliance builds for government, military, and healthcare. STIG-aligned. FedRAMP track. Dedicated engineering support.
Custom contract
<\!-- Licensing comparison table -->
Feature
Free Tier
AGPLv3
Basic · Hardcore · LLM Standard
Paid Tier
Contact us
Server · Cloud · LLM Premium · Certified
License
AGPLv3
Commercial
Use case
Individuals · Startups (pre-Series A)
Production · Enterprise · Regulated
Target hardware
Consumer PCs, laptops, EPYC/Xeon
EPYC/Xeon, native NICs, GPU (Stage 16+)
SMP / multi-core
Limited
✓ Full SMP enabled
LLM inference (CPU)
149.3 tok/s on EPYC
149.3 tok/s on EPYC
Zero LLM (built-in)
✓ Always included
✓ Always included
Zero LLM Premium
—
Anthropic / OpenAI / Gemini
Ring0 Cloud
—
Pre-installed Zero, BYO model
Compliance builds
—
Government · Military · Healthcare
Support
Community (GitHub)
Priority email + SLA
Zero Basic + Zero Hardcore are 100% free, forever. The free tier ships complete — it is not a trial. Ring0's startup program offers free Zero until Series A for qualified startups.
Benchmark

Real hardware.
Real numbers. Not a slide deck.

Zero Server — AMD EPYC 9354P
64 logical CPUs · Native NIC · SMP enabled · No GPU
149.3tok/s
Qwen3-1.7B Q4_K_M — 32-token run, CPU-only, Ring-0
System / Runtime
Model
Mode
Result
Zero Server — AMD EPYC 9354P
Qwen3-1.7B Q4_K_M
CPU-only, Ring-0
149.3 tok/s
Fast desktop CPU, llama.cpp / Ollama class
Qwen3-1.7B Q4_K_M
CPU-only
10–15 tok/s
Intel i7-13700K, llama.cpp Q4
Qwen3-1.7B
CPU-only
35 tok/s
Consumer GPU, 6 GB VRAM
Qwen3-1.7B
GPU
~126 tok/s
Qwen official SGLang benchmark, 1 GPU
Qwen3-1.7B BF16
GPU serving
227.80 tok/s

GPU acceleration is the next target. The architecture doesn't exclude it — it delays it until the Ring-0 foundation is solid. Stage 16+.

We don't sandbox the AI inside a guest.
The AI is the kernel.
Roadmap
Stage 11 — Now
Boot-LLM Operational
Cross-platform deterministic. Validated on AMD EPYC 9354P. CPU-only inference at 149.3 tok/s. AGPLv3 open source.
Stage 12
Sandboxing + AgentLang
Full sandbox isolation between agents. AgentLang migration from experimental to production runtime. Type-safe agent execution.
Stage 14
Dual Desktop — Human Oversight
Classical user desktop alongside Agent Desktop: live visualization of agent reasoning, state, and intent. Pillar 2: Human Oversight at OS level.
Stage 24+
Pure-AgentLang Kernel
The AI kernel written in AgentLang itself. Self-hosting. The OS that runs itself. The logical endpoint of what we started.
Open Source

The kernel is free.
It stays free.

Zero Basic, Zero Hardcore, and the Zero Server foundation are AGPLv3. The kernel source is published openly. Early access is open to design partners and core contributors. Ring0's startup program gives free Zero access until Series A — because the ecosystem matters more than early revenue.

Star us on GitHub GitHub stars
All Editions
Zero Basic
Everyday local AI OS. Individuals.
Free · AGPLv3
Zero Hardcore
Bare kernel. Builders and hackers.
Free · AGPLv3
Zero LLM Standard
Built-in for every Zero user.
Free
Zero LLM Premium
Anthropic / OpenAI / Gemini rental.
$/month
Zero Server
EPYC/Xeon, production tooling.
Paid
Ring0 Cloud
Server rental, pre-installed Zero.
Paid
Ring0 Benchmark
Self-hosted free · Cloud ranked.
Free / Paid
Zero Certified
Government · Military · Healthcare.
Enterprise

The next operating
system has already started.

Zero is not a roadmap. It is shipped code running on real server hardware. The kernel is open. The vision is clear. The AI era needs a new OS — and it's here.

Get Zero on GitHub
Stage 11 of 24+ — and accelerating