Those with technical expertise at the helm are essential for the rise of domestic large-scale models.

Authored by: Jǔ dàmíng ěr

Recently, a domestic AI team named "Moonshot AI," with a staggering financing of over 1 billion US dollars, has shaken the entire tech industry.

This financing not only affirms the strength of "Moonshot AI," but also heralds the gradual solidification of the "backbone" of domestic AI large-scale models.

Established in March 2023, "Moonshot AI" is led by Professor Yang Zhilin from the School of Interdisciplinary Information Studies at Tsinghua University. The team also includes talents from international tech giants such as Google, Meta, and Amazon.

Its previous round of financing in 2023 secured over 200 million US dollars from investors including Sequoia China and ZhenFund.

Following this financing, in less than a year since its establishment, "Moonshot AI" has already been valued at approximately 2.5 billion US dollars.

So, what lies behind this massive financing? What makes "Moonshot AI" so impressive that the capital market is pouring in investments?

Team Introduction

The team at Moonshot AI stands out for its youth, professionalism, and extensive experience. The founder, Yang Zhilin, although a post-90s individual, has accumulated profound academic achievements.

Not only does he hold a background in computer science from Tsinghua University and a Ph.D. from Carnegie Mellon University, but he has also collaborated with multiple Turing Award winners in academic publications. His academic accomplishments and industry experience provide strong technical endorsement for Moonshot AI.

Yang Zhilin (middle)

Similarly, other key members such as co-founders Zhou Xinyu and Wu Yuxin, with backgrounds from Tsinghua University and work experience at renowned tech companies like Megvii Technology and Meta, have participated in projects such as Google Gemini, Google Bard, Pangu NLP, and Wudao. This positions Moonshot AI with leading research and development capabilities in the field of large-scale models.

Wu Yuxin

It can be said that the strong technical background and research capabilities, as well as the diversity of the team members' backgrounds, are the reasons why various capital sources are optimistic about Moonshot AI.

Only with strong technical capabilities and the ability to approach problems from different perspectives can one better understand and meet the market's demands.

Product Introduction

Compared to some large-scale models on the market that pursue comprehensive capabilities, Moonshot AI's large-scale model Kimi Chat focuses more on the ability to process long texts.

For example, its practical usage can support approximately 200,000 Chinese characters in context, which is 2.5 times that of Anthropic's Claude-100k (measured at approximately 80,000 characters) and 8 times that of OpenAI's GPT-4-32k (measured at approximately 25,000 characters).

At the same time, through innovative network structures and engineering optimizations, Kimi Chat achieves lossless long-range attention mechanisms under a trillion parameters, without relying on "shortcut" solutions such as sliding windows, downsampling, or small models that significantly compromise performance.

These improvements allow Kimi Chat to process inputs of up to 200,000 Chinese characters without sacrificing comprehension and generation quality, which is very rare among current AI models.

This advantage also enables Kimi Chat to demonstrate significant potential in fields such as finance, law, and scientific research, where rapid analysis and summarization of lengthy documents are required.

Summary and Analysis

From both a technical and market perspective, Moonshot AI and its large-scale model Kimi Chat stand out among numerous domestic large-scale models for two main reasons:

Firstly, Kimi Chat's focus aligns with the technical essence of current large-scale models.

After all, the current AI large-scale models are essentially born to assist humans in processing overwhelming information in the era of information explosion. This is something that only entrepreneurs with profound academic backgrounds like Yang Zhilin can deeply understand.

Therefore, Kimi Chat's ability to process long texts precisely meets the market demand for "handling massive information."

Secondly, Kimi Chat's positioning in the consumer market.

While many domestic large-scale models either conservatively or blindly enter the business market, Kimi Chat is positioned in the consumer market, differentiating itself from similar large-scale models by providing personalized and convenient AI services.

Moreover, users of consumer products typically provide direct feedback on the product's experience and functionality. This immediate feedback mechanism allows Kimi Chat to quickly iterate and optimize its products, thereby maintaining a leading position in the fiercely competitive market.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。