A compact text model built on Qwen3 architecture, published by contextboxai. At 1.7B parameters with a 40K token context window, it trades raw capability for efficiency and deployability. Open weights under Apache 2.0 make it accessible for local or embedded use cases.