From the source
NVIDIA has made Nemotron 3.5 Bolt E4B 4GB available, a small open model it says is built for local assistants, interactive characters and robots that can hold a conversation and use tools.
- NVIDIA describes it as its newest small open model, positioning it for on-device work rather than cloud serving.
- In a demonstration, Reachy Mini chats and moves with Bolt handling conversation and NVIDIA Cosmos handling vision, and NVIDIA says both models run locally on Jetson Orin Nano 8GB.
- The post gives no license, context window or price for the model.
