The current window for open-source AI viability is narrowing. Nathan Lambert argues that open models face a precarious six-month survival period against proprietary dominance. This timeline forces a reckoning for developers relying on open weights. Practitioners must now decide if community-driven research can outpace the massive compute advantages held by closed-model labs.