#Installing openclaw with Local LLM procedure

1 messages · Page 1 of 1 (latest)

vocal beacon
#

Hi @celest pier ,

I amd going to install oenclaw and integrate local llm .
guide me with step by step process

hasty veldt
#

Maybe Krill was down on this day. I don't see any answers.

celest pierBOT
#

Here’s a clean step-by-step to install OpenClaw and run it on a local LLM (most reliable path = Ollama).

0) First: what OS + where will the LLM run?

Tell me:

  1. macOS / Linux / Windows (WSL2?)
  2. Will the LLM run on the same machine as the OpenClaw Gateway, or another host on your LAN?

(I’ll still give the default steps below.)

1) Install OpenClaw (recommended)

Follow the official install page: https://docs.openclaw.ai/install

macOS / Linux / WSL2:

# 
curl -fsSL https://openclaw.ai/install.sh | bash

After install, verify:

openclaw --version
openclaw doctor
openclaw status

Then run onboarding (sets up config + can install the daemon/service):

openclaw onboard --install-daemon

Open the dashboard:

# 
openclaw dashboard

2) Install a local LLM runtime (Option A: Ollama)

Provider docs: https://docs.openclaw.ai/providers/ollama

  1. Install Ollama from https://ollama.ai and make sure it’s running.
  2. Pull at least one model in Ollama (pick any you want).

Quick check that Ollama API is reachable:

curl http://localhost:11434/api/tags
#

3) Connect OpenClaw to Ollama (simple “auto-discovery” mode)

Ollama doesn’t need a real API key, but OpenClaw uses the key as an “opt-in” signal:

openclaw config set models.providers.ollama.apiKey "ollama-local"

Now set your default agent model to an Ollama model you have installed:

openclaw config set agents.defaults.model.primary "ollama/<your-ollama-model-tag>"

Restart the gateway:

# 
openclaw gateway restart

Verify OpenClaw sees models:

openclaw models list

4) If Ollama is on another machine (LAN) (important detail)

Use Ollama native URL (no /v1). The docs are explicit: https://docs.openclaw.ai/providers/ollama

In your config, set:

  • models.providers.ollama.baseUrl: "http://OLLAMA_HOST:11434" (no /v1)
  • models.providers.ollama.api: "ollama"