Set up Ollama and local models

Install and run Ollama

Confirm the local runtime is installed, running, and accessible from the same computer as LobeWork.

Choose a model that fits

Model size affects memory, storage, speed, and answer quality. Start smaller and increase only when the device can support it.

Start smaller and increase only when the device can support it.

Connect from LobeWork

Add the local provider endpoint and allow LobeWork to detect available models.

Test with a simple prompt

Confirm that the model responds before attaching large project context.

Understand privacy limits

Local inference can keep prompts on the device, but connected tools, downloads, and update checks may still require internet access.