OpenClaw VPS Specs
Minimum vs Recommended
Exactly how much RAM, CPU, and storage you need β broken down by what you’re actually doing with your agent.
RAM Is Everything
CPU and storage matter, but RAM is the single most important spec for OpenClaw. Here’s what each tier gets you.
The Gateway, Agent, and Docker runtime together consume ~1.5β2 GB at idle. Each active browser automation session (Chromium) adds 1β2 GB on top. If your VPS runs out of RAM, OpenClaw doesn’t slow down β it crashes and stops responding until you restart it.
Specs by What You’re Actually Doing
Stop guessing. Match your use case to the right server size.
All Specs at a Glance
Summary table for fast decision-making.
| Use Case | β οΈ Minimum | β Recommended | π Power |
|---|---|---|---|
| Personal assistant (1β2 channels) | 4 GB / 1 vCPU / 40 GB | 8 GB / 2 vCPU / 80 GB NVMe | β |
| Multi-channel + automations | 8 GB / 2 vCPU / 80 GB | 16 GB / 4 vCPU / 200 GB NVMe | β |
| Browser automation (Chromium) | 8 GB / 2 vCPU / 100 GB | 16 GB / 4 vCPU / 200 GB NVMe | β |
| Local AI model β 7B (Ollama) | 16 GB / 4 vCPU / 200 GB | 24 GB / 6 vCPU / 300 GB NVMe | 32 GB+ |
| Local AI model β 13B (Ollama) | 24 GB / 6 vCPU / 300 GB | 32 GB / 8 vCPU / 400 GB NVMe | 48 GB+ |
| Small team (2β5 users) | 16 GB / 4 vCPU / 200 GB | 32 GB / 8 vCPU / 400 GB NVMe | β |
OpenClaw’s Memory system reads and writes SQLite databases and Markdown files constantly. NVMe SSD is 3β5Γ faster than standard SSD for these small random I/O operations. It directly impacts how quickly your agent responds to requests. Always choose NVMe if available β Hostinger includes it on all plans.
Found Your Spec Level?
Compare hosting providers to find the best match for your required specs and budget.
Compare Hosting Providers Installation Guide β
