WebLLM · on this device only

Before you continue

This experience runs a language model directly on your device. Nothing will start until you choose to proceed.

• The first run downloads the model files and keeps them in the browser cache, using network bandwidth and local storage.
• Loading and running the model can use significant GPU or shared memory, slow down other apps, warm the device, and consume battery.
• Dialogue generation happens locally in your browser; no remote inference service receives the conversation.

You can choose a lighter model from Setup before continuing.