#install fresh openclaw via ollama
1 messages ยท Page 1 of 1 (latest)
Yep, fastest path is:
pnpm add -g openclaw@latest
pnpm approve-builds -g
openclaw onboard --install-daemon
During onboarding, choose:
- Provider:
Ollama - Base URL:
http://127.0.0.1:11434 - Mode:
Local
orCloud + Localif you also want Ollama cloud models
A few important notes:
- Install/start Ollama first: https://ollama.com/download
- You can pre-pull a model with
ollama pull <model>, but onboarding can also discover and auto-pull for you - If Ollama is on another machine, use
http://host:11434with no/v1./v1breaks tool calling
Then verify:
openclaw gateway status
openclaw dashboard
Docs:
- https://docs.openclaw.ai/start/getting-started
- https://docs.openclaw.ai/providers/ollama
- https://docs.openclaw.ai/gateway/local-models
One honest caveat: Ollama is fine for local-first setup, but local models are usually rougher for tool-heavy agent workflows unless you have pretty serious hardware.
If you want, paste these and Iโll walk you through your exact setup:
openclaw --version
node --version
ollama --version
and tell me your OS.
windows os
rtx 4070 laptop
16gb ram
With Windows, Iโd go WSL2 + Ubuntu. Native Windows works, but WSL2 is still the smoother OpenClaw path.
Your hardware is fine to try Ollama locally, but 16 GB RAM is a bit tight for strong tool-using agent workflows. Good for testing, just keep expectations realistic.
Recommended path
- In PowerShell (Admin):
wsl --install -d Ubuntu-24.04
- In WSL/Ubuntu, enable systemd:
sudo tee /etc/wsl.conf >/dev/null <<'EOF'
[boot]
systemd=true
EOF
Then back in PowerShell:
wsl --shutdown
- Reopen Ubuntu, make sure Node 24 and pnpm are installed inside WSL, then install OpenClaw:
pnpm add -g openclaw@latest
pnpm approve-builds -g
openclaw onboard --install-daemon
- During onboarding choose: