A mid-sized open-weight text model from the Qwen 3 family, published by z-lab as a bfloat16 variant with a DFlash configuration. It handles text-in, text-out tasks within a 40K token context window. Specific capability details beyond its architecture lineage are limited.