Free AI, right here.

Choose an on-device model or a free hosted model through OpenRouter.

Free · on-device
Ready. The model downloads the first time you send a message.
What can I help with?

Ask a question or try an idea.

Enter to send · Shift + Enter for a new line

First use downloads approximately 180 MB for SmolLM2 or 790 MB for Qwen from Hugging Face; a stable connection and enough device memory are needed. Downloads may use your data plan. On-device chat messages are not sent to a model provider. Hosted chat sends messages through Roven to OpenRouter and its selected provider; shared capacity is limited. Models can make mistakes.