A compact embedding model from the Qwen 3 family, this variant produces vector representations of text rather than generating language. Its small footprint and open-weight nature make it accessible for local deployment, though as an internal testing artifact its stability and documentation may be limited.