开源大模型

Search documents
后DeepSeek时代:六小虎向左,BAT向右
投中网· 2025-04-09 02:27
以下文章来源于新熵 ,作者茯神 新熵 . 洞察商业变量,探寻商业本质。 将投中网设为"星标⭐",第一时间收获最新推送 国内大模型的竞争规则变了。 来源丨 茯神 编辑丨 思原 来源丨 新熵 DeepSeek给AI大模型行业,免费赠送了一波国民级别的市场教育,却也平等地在先行者们头上,悬起了一把达摩克利斯之剑。 其中,AI"六小虎"之中的智谱就是一个缩影,智谱脱胎于清华大学知识工程研究室,素来有"国家队"之称。然而就在最近开始频频出现融资动作,10天 之内补充弹药达15亿人民币;可与此同时,组织震荡颇有加剧之势,从一线团队到高管大牛皆有波及。冰火两重天的态势,可谓是目前除了DeepSeek 之外,大多数大模型从业者们,真实写照的一个缩影。 2024年底,智谱曾以200亿元的估值,完成一轮30亿元人民币的融资,在这之后,包括杭州城投、上乘资本、华发集团等国资背景的资方快马加鞭地 赶到为其注资。 不过,也有风投人士对「新熵」分析,DeepSeek的横空出世还是对智谱的估值造成了一定负面影响,快速拿钱也可能是为了抢下已经出现上涨瓶颈的 相对高价。 与大开现金粮仓之门形成反差的是,智谱在团队规模和对外投资上呈现出收缩之势 ...
开源浪潮席卷全球,大模型亟需转型“商业化2.0”?
3 6 Ke· 2025-04-08 12:12
Core Viewpoint - The article discusses the shift towards open-source models in the AI industry, highlighting that 2025 marks a significant turning point as major tech companies embrace open-source strategies despite the initial success of closed-source models in commercialization [2][3]. Group 1: Open-source vs Closed-source - The "closed-source" camp focuses on monetization through technology protection, ensuring service quality and data security, while the "open-source" camp promotes accessibility and innovation through shared models and community collaboration [3]. - The rise of open-source models, exemplified by companies like DeepSeek, has initiated an unprecedented "open-source wave" in the global AI industry [3]. Group 2: Major Players and Their Contributions - Major tech companies have released numerous open-source models, with significant contributions from firms like OpenAI, Google, Meta, and Alibaba, showcasing advancements in model performance and capabilities [2][5][6]. - Notable releases include Meta's Llama 4, which is highlighted as one of the most advanced multi-modal models, and DeepSeek's models that have achieved top rankings in open-source performance [5][6]. Group 3: Drivers of Open-source Adoption - The article identifies four key drivers behind the surge in open-source models: the rise of edge intelligence, the need for industry-specific customization, accelerated ecological division of labor, and the crossing of a technological threshold that enhances model usability [11][12][13]. - Open-source models are seen as a means to democratize technology, reduce costs, and foster innovation among developers and small enterprises [14][15]. Group 4: Commercialization Strategies - Companies are exploring various commercialization strategies for open-source models, including offering basic models for free while charging for premium API services, creating community and enterprise versions, and leveraging cloud platforms for monetization [16][17][20]. - The trend indicates a move towards hybrid models that balance open-source initiatives with sustainable revenue generation [20].
杭州的程序员们赢麻了!一举包揽全球前三
Sou Hu Cai Jing· 2025-03-30 03:51
Core Insights - The latest trend report from Hugging Face highlights that the top three open-source AI models are all from Hangzhou, namely DeepSeek-V3-0324, SpatialLM, and Qwen2.5-Omni-7B, surpassing models from Nvidia and Google [1][3]. Group 1: Model Performance and Development - DeepSeek-V3-0324, released on March 26, is a significant upgrade from its predecessor, showing remarkable improvements in reasoning, code generation, Chinese writing, and search capabilities, outperforming Claude-3.7-Sonnet and matching the quality of Qwen-Max [3][5]. - Qwen2.5-Omni-7B, launched just 24 hours after its release, can handle multiple input types including text, images, audio, and video, achieving record-breaking performance in OmniBench evaluations [5][6]. - SpatialLM, a newcomer developed by Qunke Technology, quickly rose to the second position on the leaderboard within ten days of its release, showcasing its ability to understand and reconstruct 3D scenes from video data [7][8]. Group 2: Industry Impact and Ecosystem - The rapid development of these models reflects a competitive yet collaborative environment in Hangzhou's AI sector, with companies like DeepSeek and Alibaba pushing each other to innovate [10][14]. - The open-source approach has allowed numerous companies to leverage these advanced models at low costs, accelerating the development of vertical models and potentially transforming AI capabilities in countries with weaker AI foundations [19][20]. - The Qwen series has seen over 40 million downloads globally, indicating a strong demand and impact on the AI ecosystem [19]. Group 3: Community and Cultural Aspects - Hangzhou is characterized by a strong open-source culture, with significant contributions from local tech giants like Alibaba and emerging startups, fostering a vibrant community of developers and AI entrepreneurs [25][26]. - The establishment of the ModelScope community aims to lower the barriers to AI model usage and promote the development of China's AI ecosystem, serving over 1 million developers [26].
后DeepSeek时代:六小虎向左,BAT向右
3 6 Ke· 2025-03-25 11:23
后DeepSeek时代:六小虎向左,BAT向右 DeepSeek犹如一颗投入平静湖面的巨石,在AI行业掀起了滔天的波澜,甚至可以夸张点说,其直接改写了国内大模型的竞争规则。 DeepSeek给AI大模型行业,免费赠送了一波国民级别的市场教育,却也平等地在先行者们头上,悬起了一把达摩克利斯之剑。 其中,AI"六小虎"之中的智谱就是一个缩影,智谱脱胎于清华大学知识工程研究室,素来有"国家队"之称。然而就在最近开始频频出现融资动作,10天之 内补充弹药达15亿人民币;可与此同时,组织震荡颇有加剧之势,从一线团队到高管大牛皆有波及。冰火两重天的态势,可谓是目前除了DeepSeek之 外,大多数大模型从业者们,真实写照的一个缩影。 2024年底,智谱曾以200亿元的估值,完成一轮30亿元人民币的融资,在这之后,包括杭州城投、上乘资本、华发集团等国资背景的资方快马加鞭地赶到 为其注资。 不过,也有风投人士对「新熵」分析,DeepSeek的横空出世还是对智谱的估值造成了一定负面影响,快速拿钱也可能是为了抢下已经出现上涨瓶颈的相 对高价。 与大开现金粮仓之门形成反差的是,智谱在团队规模和对外投资上呈现出收缩之势。高峰期阶段的 ...
大模型全开源了,那到底咋挣钱啊?
虎嗅APP· 2025-03-18 09:51
Core Viewpoint - The article discusses the paradox of open-source large models in the AI industry, questioning how these models can generate revenue despite being freely available. It emphasizes that profitability is essential for business operations and suggests various monetization strategies that can be employed by companies in this space [5][8][41]. Group 1: Open Source Models and Revenue Generation - Open-source models have become mainstream, but there is skepticism about their ability to generate revenue [4][7]. - Companies can monetize open-source models through several strategies, such as charging for usage rights of certain models [12][18]. - Successful examples from the open-source world, like Red Hat, illustrate that companies can provide paid solutions around open-source products [9][10]. Group 2: Monetization Strategies - Companies can charge for customized B2B model deployments, which is a significant revenue source [20][33]. - Selling computational power, as demonstrated by DeepSeek, is another viable revenue stream, with reported daily profits of $470,000 and a profit margin of 545% [22][23]. - Open-source products often generate more revenue from services rather than direct product sales, creating an ecosystem that supports monetization [28][30]. Group 3: Market Dynamics and Challenges - The AI industry is still evolving, and many companies are struggling to achieve profitability, with significant investments in GPU resources yielding limited returns [45]. - The article highlights that the current focus for AI companies should be on gaining attention and user engagement rather than immediate profitability [47]. - The competitive landscape necessitates that companies adopt open-source strategies to remain relevant and avoid being overshadowed by leaders like DeepSeek [47][48].
杭州豪宅售楼处里,又能看见阿里员工了
阿尔法工场研究院· 2025-03-02 11:42
以下文章来源于南方地产观察 ,作者察叔 南方地产观察 . 易简集团旗下地产新媒体,大湾区高净值人群买房都看这儿。 作 者 | 察叔 来源 | 南方地产观察 导语 :阿里回来了,惨跌的未来科技城也涨了。 新年以来,阿里的股价涨疯了,港股从2025年1月13日的最低点77.35港元涨至2月28日的124.2港 元,涨幅达60%。市值从约1.6万亿港元增至2.36万亿港元,增量近8000亿港元。 原因很简单,DeepSeek带来的开源大模型,导致阿里云业务价值得到重估。 阿里股价大涨,阿里员工都很兴奋。 有阿里员工爆料,同事们都已经开始看房了,阿里巴巴集团总部在杭州市滨江区网商路,但余杭区 西溪总部园区,也是重要大本营,所以余杭区的大别墅最受阿里员工欢迎。 一位阿里员工表示,有同事正在看余杭区的郡西山墅,以前都觉得贵,买不起,现在终于有机会 了。 郡西山墅,在良渚文化村,主打山景中式合院,面积约200平-679平,随便一套都是2000万起步, 最高达半个小目标。 另一位阿里员工表示,正在看拱墅区的绿城润百合。绿城润百合主打高层和叠墅,叠墅产品面积约 240平-280平,总价1600万-2200万。 "以前阿里跌的时 ...