Grok 4 Heavy

Search documents
Grok 4长流程工作应用潜力初显 带动AI Infra与算力需求
智通财经网· 2025-07-12 07:50
智通财经APP获悉,中信证券发布研报称,Grok 4在专业学科和复杂任务上的推理能力突出,展现未来 模型在长流程专业工作上的应用潜力,支持Agent落地高价值场景,结合后续多模态能力有望突破打开 全新应用场景,行业落地对应带动AI Infra和算力需求,建议关注相关领域重点公司的投资机会,综合 梳理以下投资主线:1)主线一:通用管理软件; 2)主线二:工具软件和其他重点行业软件;3)主线 三:AI基础设施。 中信证券主要观点如下: 事项:Grok 4正式发布并开放使用 2)Vending-Bench:在衡量复杂任务解决能力的商业环境测试Vending-Bench中,Grok-4得分是第二名 Claude Opus 4的两倍,模型正朝着解决真实复杂问题的方向迈进。 3)其他:在GPQA、AIME25、HMMT 25、USAMO 25等专业学科知识测试集上,Grok 4 Heavy在其中 4项夺冠,尤其在 AIME25与HMMT25 分别获得100% /96.7%的接近满分表现。 推理能力发展带动算力需求,技术创新为后续模型推理提效带来新思路 训练侧,Grok 4较Grok 2训练量提升了100倍,较Grok-3 ...
X @Elon Musk
Elon Musk· 2025-07-10 22:49
RT Mckay Wrigley (@mckaywrigley)My thoughts on Grok 4 Heavy after 12hrs:Crazy good!“Create an animation of a crowd of people walking to form “Hello world, I am Grok” as camera changes to birds-eye.”And it 1-shotted the *entire* thing.No other model comes close.Watch the full clip. https://t.co/4j1GXNIF9O ...
年费最高超2万元!20万GPU训出Grok 4 马斯克的“野心”被质疑 木头姐:20万亿美元蛋糕正被xAI和OpenAI等瓜分
Mei Ri Jing Ji Xin Wen· 2025-07-10 14:37
当地时间7月9日,马斯克旗下xAI的下一代大模型Grok 4系列正式发布。 马斯克在发布会上强调,Grok 4是目前世界上最聪明的AI。他还称,Grok 4在所有学科上都超越了博士水平,没有例外。不过,Grok 4的订阅费相当昂贵, 最高达3000美元/年(约合人民币21530元)。 大模型性能评估平台Artificial Analysis的全套基准测试成绩表明,Grok 4已经成为当前领先的AI模型,总成绩达到了73分,领先于o3、Gemini 2.5 Pro、 Claude 4 Opus等模型。 但马斯克对于Grok 4的野心远不止于此,他此前表示,要用具有高级推理能力的Grok 4重写人类知识库,补充缺失的内容,纠正错误的知识,再基于新 的"干净而准确"的知识库重新训练AI。不过,这一说法也遭到了业界人士的质疑。 被称为"木头姐"的凯西·伍德(Cathie Wood)表示,Grok虽然起步较晚,但在性能上很快追赶上o3 pro等头部模型,这得益于训练集群的合理布局。在Grok 4亮相前一周,xAI刚刚完成新一轮百亿美元融资。截至目前,xAI累计融资额已超过200亿美元。 "世界最强AI",年费最高达30 ...
Grok 4正式发布!性能媲美GPT-5和Claude 4 Opus,史上最有“网感”的大模型?
硬AI· 2025-07-10 08:30
Grok 4拥有25.6万token的上下文窗口,主打多模态功能,支持更复杂的交互形式,同时具备更快的推理速度和改进的用 户界面。该模型订阅费为30美元/月,Heavy版本的费用为300美元/月。 硬·AI 作者 |李笑寅 编辑 | 硬 AI 当地时间9日晚,xAI公司旗下AI聊天机器人的最新版本Grok 4正式发布。 北京时间10日上午11:00,发布会直播正式开始。期间,xAI官方发推宣称, Grok 4是最新、最强大的旗 舰模型。 马斯克表示,Grok 4能做到GRE任何学科接近满分,最强大的是其推理能力,已经实现了超越人类的推理 水平。 "它几乎比所有学科的研究生都更聪明。" 据发布会介绍, Grok 4的订阅费为30美元/月,更强大的Grok 4 Heavy版本的费用为300美元/月,Grok 3维持免费开放。 时间表方面, Grok 4 API现已开放,8月将推出编程版本,9月推出多模态智能体版本,10月推出视频模 型。 此前,马斯克决定跳过Grok 3.5版本、直接发布Grok 4,这一"野心勃勃"的做法使得本次发布会备受关 注。 01 性能与GPT-5和Claude 4 Opus相媲美 据发布 ...
Grok 4强势发布!马斯克:它是在所有学科同时达到博士后水平的唯一存在
Sou Hu Cai Jing· 2025-07-10 07:11
Core Viewpoint - The release of Grok 4 by xAI marks a significant advancement in AI capabilities, with claims of achieving postdoctoral-level proficiency across multiple disciplines, potentially leading to groundbreaking scientific discoveries within the year [2][8]. Group 1: Product Details - Grok 4 is available in two subscription versions: Grok 4 at $30/month and Grok 4 Heavy at $300/month, with the latter's annual fee exceeding 20,000 RMB [4][5]. - Grok 4 Heavy scored 44.4% in the Human Last Exam (HLE), outperforming the previous top model, Gemini 2.5 Pro, which scored 26.9% [5][8]. Group 2: Performance and Testing - Grok 4 excelled in the HLE test, which spans 100 disciplines and includes 2,500 doctoral-level questions, indicating a significant breakthrough in complex knowledge systems and deep thinking capabilities [8]. - The model has achieved top scores in various prestigious tests, including HMMT, USAMO, and GPQA, and received a perfect score in the AIME25 [13][14]. Group 3: Technological Advancements - The training volume from Grok 2 to Grok 4 increased by 100 times, with enhanced training efficiency through data selection and algorithm optimization [9]. - Grok 4's reasoning ability improved by 10 times compared to its predecessor, aided by the use of the world's top supercomputing clusters and increased reinforcement learning investments [9]. Group 4: Future Developments - xAI plans to release additional models, including a coding model in August, a multi-model agent in September, and a video generation model in October, focusing on enhancing visual capabilities [19][20].
马斯克xAI发布Grok 4:训练算力提升100倍,多项测试中领先第二名一倍
Feng Huang Wang· 2025-07-10 06:20
Core Insights - xAI has launched its latest large language model, Grok 4, which shows significant performance improvements over its predecessor, Grok 3, with a 100-fold increase in training computational power [1] - Grok 4 achieved a 25% problem-solving rate in the "Humanities Last Exam" benchmark, while the multi-agent version, Grok 4 Heavy, exceeded 50% [1] - The company is focusing on enhancing multi-modal understanding capabilities and has released an API for Grok 4, supporting a context length of 256K [2] Model Performance - Grok 4 demonstrates superior reasoning capabilities in standardized tests, including GPQA and AIME, and achieved a perfect score in the Live Coding Bench test [2] - The model integrates tool usage directly into its training process, improving reliability in complex task handling [2] Commercialization Efforts - xAI has introduced a subscription service, Super Grok Heavy, allowing users to access both Grok 4 and Grok 4 Heavy [3] - The company plans to develop a dedicated programming model and initiate video generation model training using over 100,000 H200 GPUs in the coming weeks [3] - The release of Grok 4 marks a significant breakthrough in the competitive landscape of large language models, particularly in reasoning and multi-agent collaboration [3]