A compact, lightweight model built for efficiency on resource-constrained devices. At 360M parameters, it operates within tight memory and compute budgets, making it suited for on-device inference rather than server-side deployments. Expect capable text processing within its scale, though complex reasoning tasks will push its limits.