China’s Second Major Foundation Model Investment of 2025: Mianbi Secures New Round of Multi-Hundred-Million RMB Financing

 Here is the English version of the content with a “References” section added at the end:

China’s Second Major Foundation Model Investment of 2025: Mianbi Secures New Round of Multi-Hundred-Million RMB Financing

Mianbi Intelligence has recently completed a new financing round worth several hundred million RMB, marking the second major funding event in China’s large model sector in 2025. The round was co-led by Loongson Capital, CDH Baifu, Zhongguancun Science City Fund, and SAIF Partners. Additional investors included the Beijing Artificial Intelligence Industry Investment Fund and Qiming Venture Partners, with Wanjia Capital serving as the exclusive financial advisor.

Company Background and Technology Path

Founded in August 2022 and headquartered in Beijing, Mianbi Intelligence’s core team originates from Tsinghua University’s Natural Language Processing Lab, along with talent from leading tech firms such as Alibaba, ByteDance, and Baidu. The company is guided by the principle of “efficiency as the first principle,” focusing on innovations in foundation models and their industrial applications, especially in edge-side AI deployment.

Its flagship product line, “MiniCPM,” is a series of highly efficient on-device large models with just 2 billion parameters. Since its debut in February 2024, MiniCPM has achieved over 3 million downloads and consistently ranked on top of GitHub and HuggingFace trending charts, making it a star in the open-source community.

Commercial Strategy and Partnerships

Mianbi has rapidly advanced its commercial strategy in edge AI, establishing strong partnerships with industry leaders such as Huawei, MediaTek, Lenovo, Intel, Great Wall Motors, and Yeelight. Its business spans AI-powered smartphones, AI PCs, smart cockpits, smart homes, and embodied robots. The company also formed strategic alliances with Huawei Cloud and Baidu AI Cloud, aiming to develop edge-cloud collaborative paradigms. Their focus lies in overcoming critical challenges at the intersection of computing power and model optimization through high-knowledge-density models, improved hardware utilization, and advanced scheduling algorithms.

Industry Applications and National Projects

In vertical applications, Mianbi is deeply embedding its large model technologies in fields like law and education. In July 2024, it supported the Shenzhen Intermediate People's Court in launching the nation’s first legal-domain vertical large model, covering all common civil and commercial case types. In November 2024, Mianbi participated in the development of the "Faxin Legal Foundation Model" released by the Supreme People’s Court, designed as a national-level AI infrastructure for legal systems, enhancing judicial efficiency and leveraging legal big data to modernize court operations.

CEO's Perspective and Future Vision

CEO Li Dahai stated in an internal letter that the foundation model sector is shifting from the “Great Leap” phase to a “real combat” era. The company aims to accelerate its high-efficiency, edge-focused foundation model commercialization by delivering models that offer equivalent parameter counts but superior performance, lower power consumption, and faster processing speeds. He emphasized that Mianbi succeeded by doing three key things right: anticipating technical and product trends, applying advanced technology to vertical industries, and establishing its own “Mianbi Law” and “Density Law” to scientifically achieve breakthrough product performance.

With strong technological innovation and rapid industrial integration, Mianbi Intelligence has become a key player in China’s foundation model landscape. This latest financing round will further enhance its presence across sectors, fostering AI-driven transformation across a variety of industries.

References

  1. 爱否科技. “面壁智能完成新一轮数亿元融资，MiniCPM下载超300万.” IT之家. https://www.ithome.com/0/816/872.htm

  2. 新浪科技. “面壁智能宣布完成数亿元融资，多家知名机构投资.” 新浪财经. https://finance.sina.com.cn/tech/roll/2024-12-14/doc-inczikpt5802541.shtml

  3. 新浪财经. “CEO李大海发全员信：大模型进入实战阶段，面壁走在前列.” 新浪财经. https://finance.sina.com.cn/cj/2024-12-11/doc-incyzuvz1035698.shtml

  4. InfoQ中文网. “面壁智能李大海：我们正在用科学方法做越级超越.” InfoQ. https://www.infoq.cn/article/3EKbCSkKxDEsVkVbjPr5

评论

发表评论

此博客中的热门博文

Why Is an AI Meeting Notes App Worth $250 Million? How Granola Stands Out in a Crowded Market

图片
  In June 2025, AI meeting notes app Granola completed a $43 million Series B funding round, boosting its valuation to $250 million. Developed by a Silicon Valley startup, Granola has quickly gained attention from investors and users worldwide with its unique user collaboration model and highly personalized feature set—emerging as a standout player in the competitive AI note-taking market. 1. Granola’s Core Edge: Seamless Collaboration Between AI and Humans Granola is not just another meeting transcription tool. Instead, it enhances users' note-taking capabilities through AI assistance. Users only need to jot down key points, and Granola automatically fills in the details, generating structured, comprehensive meeting records. This “AI-assisted human” approach improves efficiency and allows users to focus on the core content of the meeting. What makes Granola especially powerful is its customization based on meeting type—whether it's a sales call, job interview, or investor pitc...
阅读全文

AI Text Generation API Documentation

图片
  1. Overview This document outlines the usage and specifications of the AI Text Generation API. Powered by advanced artificial intelligence models, this API can generate high-quality text based on user-provided prompts. It is ideal for use cases such as intelligent writing assistants, content creation tools, and chatbots. 2. Basic Information API Name : AI Text Generation API Version : 1.0 HTTP Method : POST Endpoint :  https://api.example.com/ai/text/generate 3. Request Parameters Parameter Type Required Description prompt string Yes Input prompt to guide the text generation max_tokens int No Maximum length of generated text, default is 100 temperature float No Controls randomness of output; range: 0–1, default is 0.7 4. Response Parameters Parameter Type Description code int Response status code (200 = success) message string Response message, empty on success data object Contains the generated text - text string The generated text content 5. Request Example Copy Copy { ...
阅读全文

Beyond Trial-and-Error: A New Framework for Efficient LLM Selection, Backed by ICML 2025

图片
In the era of ever-growing large language models (LLMs), choosing the right model is harder—and more critical—than ever. With countless options like GPT, LLaMA, Mistral, and DeepSeek, engineers and researchers often face the same dilemma: How do we select the best-performing model with limited resources and minimal tuning cost? A new framework from Virginia Tech may have just cracked the code. Introducing LensLLM , a model selection method that dramatically improves performance while cutting cost by nearly 90%. This groundbreaking work has been accepted to ICML 2025 , signaling a major shift in how we approach LLM deployment. From Theory to Practice: Phase Transitions in LLM Fine-Tuning Most current selection methods are based on intuition, manual tuning, or guesswork. They’re fragile, expensive, and slow. LensLLM changes that, offering a principled approach rooted in learning theory. At the core is a new derivation of PAC-Bayes generalization bounds, which mathematically reveals ...
阅读全文