Switching Models in OpenClaw

Sometimes it's necessary to switch to different models to test performance and functionality. The following demonstrates how to switch local models using Ollama.

1. Download the Model

Open the terminal and enter the following command to download the model you want to test. Here, we take ollama downloading qwen3.5:9b as an example:

xxxxxxxxxx
ollama pull qwen3.5:9b

2. Set the Default Model

Run the command to set the default model:

xxxxxxxxxx
openclaw models set ollama/qwen3.5:9b

3. Restart the OpenClaw Gateway

First, stop the gateway:

xxxxxxxxxx
openclaw gateway Stop

Restart the gateway

xxxxxxxxxx
OpenClaw gateway restart

4. Check Default Model Configuration

Check the default model settings for OpenClaw. If you see "ollama/qwen3.5:9b" displayed under "Default," the setting is successful.

xxxxxxxxxx
OpenClaw models status

5. Check Individual Plugin Model Configuration

Generally, if a default model is set, it will be automatically used during interaction. However, sometimes if a plugin model is specified in WebChat, it will not automatically switch to the default model. In this case, you need to manually change the model to the default model.

First, open WebChat.

xxxxxxxxxx
openclaw dashboard