Last released Sep 9, 2026
Validated GPU VRAM and max-context planner for modern open-weight LLM deployment.