Released under the Apache 2.0 license, this compact model is designed for long-context tasks and instruction-following scenarios, striking a balance between efficiency, transparency, and performance.