A compact, efficient model that punches above its weight for its parameter count. Running in FP8 quantization keeps memory footprint lean without dramatic quality loss, making it practical for resource-constrained deployments. It handles straightforward text tasks reliably, though complex multi-step reasoning may expose the limits of its smaller scale.