DeepSeek的推出并非只是一个简单的技术更新，它在根本上重新定义了AI模型的经济模型。在华尔街，由于对这一低成本技术可能冲击市场需求的担忧，美股科技股随之出现了大幅抛售，英伟达一夜之间的市值蒸发接近6000亿美元，这一变迁无疑让投资者对于未来的科技股走向充满了警惕与思考。

大约一周前，DeepSeek 发布了 DeepSeek-R1， 这是一款性能对标 OpenAI o1 的杰出模型，且以 MIT 许可协议开放权重。 通过 Qwen（我的团队已使用数月）、Kimi、InternVL 和 DeepSeek ...

受美国 AI 芯片禁令影响，DeepSeek 团队不得不在性能较低的 H800 GPU（而非 H100）上进行多项优化创新，最终以低于 600 万美元的计算成本完成了模型训练（研发成本不计）。

In March 2013, a 1.6-liter turbocharged midsized sedan called Qoros 3 debuted at the Geneva International Motor Show in Switzerland, grabbing the attention of a few reporters looking for something new ...

Dazzling folk performances fill the streets with a festive atmosphere in Taiyuan Ancient County, northern China's Shanxi Province. The most eye-catching spectacle is the renowned folk performance of " ...

What’s it like to celebrate the New Year in Shenzhen, a "city of immigrants"? This Spring Festival, I was invited by my aunt’s family and drove with my parents to Shenzhen for this grand holiday. With ...