2025年，是月之暗面全面开源的一年，这帮助其在社区层面获得开发者声量与生态势能，并提前建立了技术影响力。
Also: You can turn off Gemini in Gmail, Photos, Chrome, and more - here's how
，这一点在搜狗输入法2026中也有详细论述
不是因为算力不重要，而是模型和模型之间的差距，正在以肉眼可见的速度收窄。大模型之间当然有差异，但对于绝大多数企业的实际需求来说，它们已经"够用了"。当"够用"成为基准线，比拼谁的模型更聪明就变成了一场没有终点的消耗战，边际的改善却极为有限。
Trained — weights learned from data by any training algorithm (SGD, Adam, evolutionary search, etc.). The algorithm must be generic — it should work with any model and dataset, not just this specific problem. This encourages creative ideas around data format, tokenization, curriculum learning, and architecture search.
We did of course have a countdown first, because people love countdowns, because cricket is basically one unceasing countdown, an endless pencil stub ticking off names and numbers. There were 99 members of the supporting cast to be ushered to their spots, the non-Shanes of history, meat in the Ashes room.