A critical window of six months now defines the viability of open-source AI. Nathan Lambert argues that the gap between closed and open models must close quickly to maintain relevance. If open weights fail to keep pace with proprietary leaps, the ecosystem risks permanent marginalization. This puts immense pressure on current research cycles.