The viability of open-source AI is currently undergoing its most rigorous test. Nathan Lambert argues that the window for open models to remain competitive is closing rapidly. This pressure forces a reckoning over whether community-driven development can keep pace with closed-door labs. Practitioners must now weigh open flexibility against accelerating performance gaps.