OpenClaw System Requirements (2026) — RAM, CPU, GPU & Hardware Specs

Clawpedia · For Humans

OpenClaw system requirements: minimum and recommended RAM, CPU, disk and GPU for cloud APIs, local LLMs (7B-70B), self-hosting, Raspberry Pi and Windows/WSL2.

TL;DR — OpenClaw system requirements (2026): with a cloud model (OpenAI, Anthropic, Groq) you need Node.js 18+, ~500 MB disk and 2 GB RAM on any Linux/macOS/Windows machine. To run local LLMs via Ollama, plan for 8 GB RAM per 7B model, 16 GB per 13B and a GPU with 8+ GB VRAM for real-time speed.

OpenClaw is deliberately lightweight. The runtime itself is a Node.js process, so 95% of the hardware footprint comes from the AI model you choose — not from OpenClaw. This guide lists every OpenClaw system requirement for 2026: minimum specs, recommended specs, GPU needs for local models, supported operating systems, ports, dependencies and quick commands to verify your setup.

---

Minimum vs. Recommended Hardware

ComponentMinimum (cloud AI)RecommendedLocal LLMs (7B–13B)Local LLMs (34B–70B)
CPU2 cores, 1.5 GHz4+ cores, 2.5 GHz8+ cores12+ cores
RAM2 GB8 GB16 GB32–64 GB
Disk1 GB free10 GB20 GB50+ GB
GPUNot neededNot needed8 GB VRAM24–48 GB VRAM
Network1 Mbps10+ MbpsOptional (offline OK)Optional

In simple terms: if you use cloud models, almost any laptop from the last 5 years works. Local models are what push the requirements up.

---

Does OpenClaw Need a GPU?

No — not for cloud providers. OpenClaw only needs a GPU when you run models locally through Ollama, llama.cpp or vLLM. VRAM is the constraint, not raw compute:

Model sizeVRAM neededExample modelsTokens/sec (RTX 4090)
3B3 GBPhi-3 Mini, Gemma 2B~120
7–8B6–8 GBLlama 3.1 8B, Mistral 7B, Qwen 2.5 7B~85
13B10 GBCodeLlama 13B~55
34B24 GBCodeLlama 34B, Qwen 2.5 32B~25
70B48 GB (or 2× 24 GB)Llama 3.1 70B~12

Apple Silicon (M2/M3/M4) is a strong alternative to NVIDIA thanks to unified memory — an M2 Max with 64 GB runs 70B models locally without a discrete GPU.

---

Supported Operating Systems

OSStatusNotes
Ubuntu 22.04 / 24.04Fully supportedRecommended for servers and VPS
Debian 12+Fully supportedStable, long support cycle
macOS 13+ (Ventura, Sonoma, Sequoia)Fully supportedBest experience for local LLMs
Windows 11 (WSL2)Fully supportedUse WSL, not native Windows
Fedora 38+ / ArchCommunity supportedTested by contributors
Raspberry Pi OS (64-bit)SupportedCloud models only, Pi 4 4 GB+
Windows (native)ExperimentalPrefer WSL2 or Docker

---

Required Software Dependencies

SoftwareMinimumCheckPurpose
Node.jsv18.0node --versionRuntime (v20 LTS recommended)
npmv9.0npm --versionPackage manager

Optional Dependencies

Gitv2.30git --versionInstalling skills and updates
SoftwareWhen you need it
Python 3.10+Community skills written in Python
Docker 20.10+Container deployment or isolated skills
Redis 7+High-performance shared memory backend
PostgreSQL 14+Scalable memory backend for production
Ollama 0.1+Local LLM inference
FFmpeg 5+Voice interface, audio and video skills
Chromium / PlaywrightBrowser automation skills

---

Checking Your System

One-Command Health Check


openclaw doctor

Sample output:


━━━ Required ━━━
✅ Node.js: v20.11.0 (min v18)
✅ npm:     v10.2.4  (min v9)
✅ Git:     v2.43.0  (min v2.30)
✅ Disk:    45 GB free
✅ RAM:     16 GB total, 8 GB free

━━━ Optional ━━━
✅ Python:  3.11.7
✅ Docker:  24.0.7
⚠️  Redis:  not installed (needed for production memory)
⚠️  Ollama: not installed (needed for local LLMs)

---

Installing Dependencies

Node.js


# Linux (Ubuntu/Debian)
curl -fsSL https://deb.nodesource.com/setup_20.x | sudo -E bash -
sudo apt-get install -y nodejs

# macOS
brew install node

# Windows
winget install OpenJS.NodeJS.LTS

Ollama (for local LLMs)


# Linux / macOS
curl -fsSL https://ollama.ai/install.sh | sh

# Pull a model
ollama pull llama3.1:8b

FFmpeg (for voice / media skills)


sudo apt install ffmpeg       # Linux
brew install ffmpeg           # macOS
winget install FFmpeg         # Windows

---

Network & Port Requirements

PortPurposeNeeded when
443HTTPS outbound (cloud APIs)Always with cloud models
8080Webhook listenerExternal integrations
11434Ollama APILocal models
6379RedisRedis memory backend
5432PostgreSQLPostgres memory backend

Minimum firewall rule for a stock install:


sudo ufw allow out 443/tcp

---

Resource Usage by Feature

FeatureExtra RAMExtra DiskCPU impact
Core agent100 MB200 MBLow
Memory system50–200 MB1–5 GBLow
Each chat platform30 MBMinimalLow
Each active skill10–50 MBVariesLow–Med
Local LLM 7B4–8 GB4–8 GBHigh
Local LLM 13B8–16 GB8–13 GBHigh
Voice processing200 MB500 MBMedium
Web automation300 MB200 MBMedium

---

Recommended Setups by Use Case

Use caseHardwareModelMonthly cost
Budget / hobbyAny laptopGroq (free tier)$0
Privacy-first16 GB laptop or M2Llama 3.1 8B local$0
Power user32 GB + RTX 4090Llama 3.1 70B local$0
Always-on assistantRaspberry Pi 4GPT-4o mini (cloud)~$2
Small businessVPS 4 vCPU / 8 GBGPT-4o or Claude Sonnet~$30

---

Troubleshooting

ProblemFix
node: command not foundInstall Node.js 18+ via NodeSource or nvm
EACCES: permission deniedUse nvm instead of a system-wide Node install
Slow responses on local LLMModel is CPU-bound — add a GPU or drop to a smaller model
Out of memory during installFree 2 GB RAM or add swap: sudo fallocate -l 2G /swapfile
Ollama not detected by doctorEnsure it runs on port 11434: curl localhost:11434
Missing build tools (Linux)sudo apt install build-essential

---

FAQ — OpenClaw System Requirements

How much RAM does OpenClaw need?

2 GB is enough for cloud-based setups. Plan for 8 GB if you run a 7B local model, 16 GB for 13B, and 32–64 GB for 34B–70B models.

Can OpenClaw run on a Raspberry Pi?

Yes — a Raspberry Pi 4 (4 GB or 8 GB) runs OpenClaw cleanly with cloud models. Local LLM inference on a Pi is possible only for tiny 1–3B models and stays impractical.

Does OpenClaw work on Windows?

Yes, through WSL2. Native Windows is experimental — WSL2 or Docker Desktop is the supported path.

Do I need a GPU to use OpenClaw?

No. A GPU is only required if you run local LLMs. Cloud providers (OpenAI, Anthropic, Groq, Gemini) do all the heavy lifting server-side.

Which Node.js version is required?

Node.js 18 or newer. Node 20 LTS is recommended for the best long-term compatibility with skills.

How much disk space does OpenClaw use?

The core install is under 500 MB. Local models add 2–40 GB each, depending on size and quantization.

Is Docker required?

No. Docker is optional and used for isolated deployments. Bare-metal installs with npm i -g openclaw work the same way.

---

Once your machine meets these OpenClaw system requirements, continue with Setting your OpenClaw AI model and provider or check Understanding OpenClaw's Memory System to plan a production setup.

Related Articles