随着新模型DeepSeek在拥抱脸上介绍时，AI令牌篮子下降了12％

随着通用加密市场开始在红色的一周开始，AI部门面临着特别灰色的星期一早晨。 Coingecko上的AI代币篮子下跌了12％

A new AI model on the market, DeepSeek, has surged in popularity in the US since its release on 10 January, and currently sits at the #1 spot in Apple's App Store, according to app data research firm Sensor Tower.

根据App Data Research Sensor Tower的说法，自1月10日发行以来，市场上的一种新的AI模型DeepSeek在美国发行以来一直在美国受欢迎程度，目前位于Apple App Store的第一名。

DeepSeek is a large multi-modal model trained by a small Hangzhou-based startup founded in 2023 by High-flyer, billionaire Liang Wenfeng's hedge fund. Its AI engine is powered by Nvidia's H800 chips for training and its team comprises Chinese AI experts and PhD grads from ByteDance and Tencent.

DeepSeek是一种大型的多模式模型，该模型是由一家位于Hangzhou的小型初创公司训练的，该初创公司由High-Flyer，亿万富翁Liang Wenfeng的对冲基金于2023年成立。它的AI引擎由NVIDIA的H800芯片提供动力，其团队由中国的AI专家和博士学位毕业生，来自Bytedance and Tencent。

The model, which is available for free on Hugging Face, was created with just $6 million and minimal GPU resources (compared to, say, ChatGPT's $157 billion valuation), and yet DeepSeek vastly outperforms the likes of leading AI giant, OpenAI.

该模型可在拥抱面前免费使用，只有600万美元和最少的GPU资源（例如，与Chatgpt的1570亿美元估值相比），但DeepSeek的表现极大地胜过领先的AI巨头Openai。

DeepSeek’s ability to run on mobile devices further underscores its efficiency.

DeepSeek在移动设备上运行的能力进一步强调了其效率。

Wow DeepSeek R1 version 1.5B runs perfectly locally on my phone ?So you can have a model that outperforms GPT-4o and Claude 3.5 Sonnet on math in your pocket.

WOW DeepSeek R1版本1.5B在我的手机上在本地完美运行。因此，您可以在口袋里的数学上的型号优于GPT-4O和Claude 3.5十四行诗。

DeepSeek is powered by the DeepSeek-V3 model, which its creators say "tops the leaderboard among open-source models and rivals the most advanced closed-source models globally."

DeepSeek由DeepSeek-V3模型提供动力，其创建者说：“在开源模型中排名第一，并与全球最先进的封闭源型号竞争。”

DeepSeek's Surprising Success Story

DeepSeek令人惊讶的成功故事

DeepSeek has taken the US by storm, surging in popularity and quickly becoming one of the top-grossing apps in the country. The model's success is due in part to its impressive capabilities, which users have praised for being superior to other AI models on the market.

DeepSeek席卷了美国，迅速流行，并迅速成为该国最受欢迎的应用程序之一。该模型的成功部分是由于其令人印象深刻的能力，用户赞扬了它优于市场上其他AI模型。

DeepSeek is also able to run on mobile devices, thanks to its efficient design, which further contributes to its popularity among users.

DeepSeek还能够在移动设备上运行，这要归功于其高效设计，这进一步有助于其在用户中的受欢迎程度。

As a result of DeepSeek's success, crypto and Web3 companies are now flocking to partner with the model's creators. Last week, the venture capital arm of Binance, once known as Binance Labs, announced it is rebranding as YZi Labs to shift from a purely crypto-focused mandate to an ambitious strategy encompassing AI, biotechnology, and Web3 innovations.

由于DeepSeek的成功，Crypto和Web3公司现在正在与模型的创建者合作。上周，曾经被称为Binance Labs的风险投资部门宣布，随着Yzi Labs的重新命名，将从纯粹以加密货币为中心的授权转变为涵盖AI，生物技术和Web3创新的雄心勃勃的策略。

“By venturing into new horizons in AI and biotech, while staying rooted in our Web3 foundations, we aim to push the boundaries of innovation and create lasting value for society,” explained Ella Zhang, the founding head of Binance LAbs.

Binance Labs的创始负责人Ella Zhang解释说：“通过在AI和Biotech的新视野中冒险，同时扎根于我们的Web3基础，我们旨在突破创新的界限并为社会创造持久的价值。”

Meanwhile, crypto miner Riot Platforms announced it is shifting its focus away from Bitcoin to AI in an effort to maximize its unused power. Riot revealed last week that it is exploring the potential of utilizing approximately 600 megawatts (MW) of unused power capacity for AI and high-performance computing (HPC) applications.

同时，加密矿工Riot平台宣布将其重点从比特币转移到AI，以最大程度地发挥其未使用的力量。 Riot上周透露，它正在探索利用AI和高性能计算（HPC）应用程序的大约600兆瓦（MW）的潜力。

Riot has subsequently paused its Phase II Bitcoin mining expansion at its Corsicana site, which was originally expected to add another 600 MW of mining capacity.

随后，Riot在其Corsicana地点停止了其II期比特币采矿的扩张，最初预计将增加600 MW的采矿能力。