A compact, open-weight model that punches reasonably well for its size, handling text tasks with a 128K context window that lets it work through lengthy documents without losing the thread. Being NousResearch's release of Meta's Llama 3.1 8B base, it carries the efficiency trade-offs typical of smaller models — capable across general language tasks but less reliable on complex reasoning than its larger siblings. Its open-weight nature means it runs locally and can be fine-tuned.