Grok 4 Code

Search documents
X @Elon Musk
Elon Musk· 2025-08-10 11:32
RT Tech Dev Notes (@techdevnotes)Things we are looking forward in 2025 with Grok that's currently known- Native Video input- Grok 4.20- Grok 4 Code- Grok Mac App- Baby Grok- Big Video Generation Model (~October)- New Image Generation Model- Code Editor (unconfirmed)- New AI Companions- DLCs for Companions- Multi-Modal Agent (~September)- Macrohard- Grok 5 ...
估值2000亿美元?马斯克否认xAI融资
Bei Jing Shang Bao· 2025-07-13 15:27
Core Insights - xAI, led by Elon Musk, is reportedly negotiating a new funding round with a potential valuation of up to $200 billion, following a recent $10 billion capital raise split between debt and equity [1][5] - Musk has publicly denied these funding rumors, asserting that xAI has sufficient capital [1] - The company recently launched Grok 4, touted as the "world's strongest AI model," which includes significant updates and new features [1][2] Funding and Valuation - xAI's recent funding efforts include raising $10 billion, with plans to raise an additional $10 billion [1] - The company's valuation during the B round in May 2024 was only $18 billion, indicating a substantial increase in interest and valuation in a short period [1] - The upcoming funding round could mark xAI's third major financing event within two months [1] Product Development - Grok 4 is the fourth major iteration of xAI's AI models, featuring two versions: Grok 4 and Grok 4 Heavy, with the latter supporting multiple agents and a context window of up to 256k tokens [1][2] - Internal benchmarks show Grok 4 achieving approximately 25% accuracy on a comprehensive exam, comparable to OpenAI's tools [2] - The model introduces new voice modes and a programming model aimed at enhancing developer efficiency, addressing competition from GPT-4 and Meta Gemini [2][3] Controversies and Challenges - Grok 4's launch has been met with criticism due to past issues with the previous model, Grok 3, which faced backlash for inappropriate content [3] - Concerns have been raised regarding the need for more structured safety mechanisms and transparency in training methods for Grok 4 [3] - Musk envisions Grok 4 evolving into a truth-seeking AI, potentially integrating with humanoid robots to discover new technologies [3] Company Background and Market Position - xAI was founded in July 2023, with a team comprising members from notable organizations like Google DeepMind and OpenAI [3] - The company has released several models since its inception, continuously improving capabilities in reasoning, context length, and image understanding [4] - Despite political controversies surrounding Musk, investor interest in xAI and related ventures remains strong, as evidenced by SpaceX's plans to sell $1 billion in shares, potentially valuing the company at $400 billion [5]
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
Core Viewpoint - The release of Grok 4 by xAI marks a significant advancement in AI capabilities, featuring enhanced reasoning abilities and multi-modal functionalities, positioning it as a strong competitor in the AI landscape [2][4][6]. Group 1: Product Features - Grok 4 has a context window of 256,000 tokens, supporting more complex interactions and faster reasoning speeds compared to its predecessors [7]. - The subscription fee for Grok 4 is set at $30 per month, while the more advanced Grok 4 Heavy version costs $300 per month [5]. - Grok 4's reasoning ability has improved tenfold compared to earlier versions, making it capable of outperforming human reasoning in various subjects [4][7]. Group 2: Performance Metrics - Grok 4's performance is expected to be on par with GPT-5 and Claude 4 Opus, achieving top scores in benchmark tests against models like GPT-3, Gemini 2.5 Pro, and Claude 4 Opus [8]. - In the "Humanity's Last Exam" benchmark, Grok 4 achieved an accuracy rate of 26.9% in pure autonomous reasoning, setting a new industry record [10][11]. - However, Grok 4 scored only 16 points on the AGI-ARC-2 advanced reasoning test, indicating room for improvement in more challenging intelligence assessments [13]. Group 3: Market Position and Strategy - The launch of Grok 4 coincides with xAI's transformation period, following its merger with X to better develop and distribute Grok to a wider user base [19]. - Grok 4 includes a DeepSearch feature that allows it to extract real-time data from the internet, enhancing its understanding of internet culture, memes, and humor [15][16]. - The model is designed to attract "super users" seeking real-time search capabilities and intelligent coding support, similar to tools like GitHub Copilot [17]. Group 4: Controversies and Challenges - The release has been overshadowed by controversies regarding the platform's unfiltered "freedom of speech" approach, raising questions about its suitability for human interaction [20]. - The resignation of X's CEO Linda Yaccarino shortly before the launch has added to the uncertainty surrounding the platform [21]. - Some users have expressed concerns that Grok 4's technical achievements are being overlooked due to recent online controversies [22].
Grok 4正式发布!当世最强算力训练,对垒GPT-5和Claude 4 Opus
Hua Er Jie Jian Wen· 2025-07-10 07:49
Core Insights - Grok 4 is the latest flagship model from xAI, featuring a context window of 256,000 tokens, enhanced reasoning capabilities, and a multi-modal function that supports complex interactions [1][3][5] - The subscription fee for Grok 4 is set at $30 per month, while the Heavy version costs $300 per month, with Grok 3 remaining free [3][4] - Grok 4 has been trained on xAI's Colossus supercomputer, boasting reasoning abilities that surpass those of most graduate students [3][5] Performance and Features - Grok 4's reasoning capability has improved tenfold compared to its predecessor, and it is expected to perform comparably to GPT-5 and Claude 4 Opus [5][6] - In benchmark tests against models like GPT-3, Gemini 2.5 Pro, and Claude 4 Opus, Grok 4 consistently scored among the top [7] - The model achieved a record accuracy of 26.9% in the Humanity's Last Exam benchmark test, indicating strong autonomous reasoning [9] Development Timeline - The Grok 4 API is now available, with a programming version set to launch in August, a multi-modal agent version in September, and a video model in October [4] Market Position and Controversies - Grok 4 includes a DeepSearch feature that allows it to extract real-time data from the internet, enhancing its understanding of internet culture [12] - The release of Grok 4 coincides with a critical transition period for xAI, following its merger with X in March [12] - The model has faced controversy over its "free speech" approach, raising questions about its suitability for human interaction [13]
AI日报丨上调英伟达目标价!花旗银行:看好主权AI需求激增
Xin Lang Cai Jing· 2025-07-08 11:38
Group 1 - Apple's AI model executive Ruoming Pang is leaving for Meta Platforms Inc., marking a setback for Apple's AI efforts as Meta continues to aggressively recruit AI leaders with high salaries [3] - Citigroup has raised Nvidia's target price to $190 per share, citing significant growth in sovereign nations' demand for AI infrastructure, which is expected to drive Nvidia's expansion [3][4] - Nvidia's market capitalization is approaching $4 trillion, with its stock price increasing by 12% over the past month [3] Group 2 - Samsung Electronics reported a 56% year-over-year drop in operating profit to 4.6 trillion KRW (approximately $3.3 billion), exceeding analyst expectations of a 41% decline [4][5] - The decline in Samsung's performance is attributed to challenges in the AI chip sector, particularly the delay in certification of its advanced HBM3E chips by Nvidia, allowing competitors like SK Hynix to gain market share [5] - Samsung's chip division is expected to report an operating profit of 2.7 trillion KRW, which, while an improvement from the previous quarter, is still below last year's 6.5 trillion KRW [5] Group 3 - Elon Musk's AI startup xAI plans to launch its latest model Grok 4, which aims to compete with OpenAI's GPT-4o and Google's Gemini 2.5 Pro [6] - Wedbush analysts indicate that Nvidia's B200 chip supply is lagging behind demand, suggesting further growth potential in the coming quarters [7] - The demand for hard disk drives (HDD) is expected to benefit companies like Western Digital and Seagate Technology due to limited production capacity, allowing for price and margin increases [8]
AI日报丨上调英伟达目标价!花旗银行:看好主权AI需求激增
美股研究社· 2025-07-08 10:45
Core Insights - The rapid development of artificial intelligence (AI) technology is creating extensive opportunities in the market [1] Group 1: Company Developments - Apple's AI model executive Ruoming Pang is leaving for Meta Platforms Inc., which poses a setback for Apple's AI efforts [3][4] - Meta is aggressively hiring AI leaders, offering substantial compensation packages to attract talent [4] - Samsung Electronics reported a significant 56% year-over-year drop in operating profit to 4.6 trillion KRW (approximately 3.3 billion USD), attributed to inventory pressures and setbacks in AI chip development [9] Group 2: Market Trends - Citigroup raised Nvidia's target price to 190 USD per share, citing increased demand for AI infrastructure from sovereign nations, indicating potential for further expansion [5][6] - Nvidia's market capitalization is approaching 4 trillion USD, with a 12% increase in stock price over the past month [6] - Analysts predict that sovereign demand could contribute billions in revenue by 2025, with expectations for further growth by 2026 [8] Group 3: Product Launches and Supply Chain Insights - Tesla's AI startup xAI plans to launch Grok 4, a new AI model, which aims to compete with OpenAI's GPT-4o and Google's Gemini 2.5 Pro [11][12] - Wedbush analysts noted that demand for Nvidia's B200 chips exceeds supply, suggesting growth potential in the coming quarters [13] - The hard disk drive market is expected to benefit Western Digital and Seagate due to limited production capacity, allowing for price and margin increases [15]
马斯克再掀大模型风暴!xAI 计划周三发布 Grok-4 特斯拉 FSD 与 Optimus 迎来 “AI 超级引擎”?
贝塔投资智库· 2025-07-08 04:14
Core Viewpoint - xAI, founded by Elon Musk, is set to release its latest AI model, Grok-4, which is expected to significantly enhance Tesla's Full Self-Driving (FSD) and Optimus humanoid robot capabilities [1][4]. Group 1: Grok-4 Release and Features - Grok-4 is anticipated to outperform OpenAI and Google's latest AI models in language, mathematics, and reasoning [2]. - Grok-4 Code will serve as a personal coding assistant for top developers, potentially integrating into customized code editors [2]. - The release of Grok-4 is expected to intensify competition with OpenAI's GPT series and Google's Gemini 2.5 Pro [2]. Group 2: AI Application Trends - The emergence of low-cost AI training and inference paradigms is likely to lead to a surge in generative AI applications across various industries, boosting productivity [3]. - Companies like C3.ai, APPlovin, and Palantir have reported strong performance and outlooks, indicating robust demand for enterprise AI applications [3]. Group 3: Integration with Tesla's AI Systems - The integration of Grok with Tesla's AI supercomputing system is expected to enhance the reasoning capabilities of FSD and Optimus, likened to a leap in human intelligence [4][5]. - Grok's capabilities in natural language understanding, reasoning, and physical simulation will complement Tesla's real-time data processing and decision-making abilities [4]. - The combined system will enable simultaneous task handling in autonomous driving, enhancing obstacle detection, path planning, and in-car voice interaction [5]. Group 4: Market Reaction - Following Musk's announcement of forming a new political party, Tesla's stock fell nearly 7%, resulting in a market cap loss of over $68 billion, raising concerns about the company's core automotive business [5].
马斯克再掀大模型风暴! xAI计划周三发布Grok-4 特斯拉FSD与Optimus迎来“AI超级引擎”?
智通财经网· 2025-07-08 01:23
Core Insights - Tesla's CEO Elon Musk announced the upcoming release of the AI model Grok-4 by his startup xAI, which is expected to significantly enhance Tesla's Full Self-Driving (FSD) and Optimus humanoid robot capabilities [1][2] - Grok-4 is anticipated to outperform OpenAI and Google's latest AI models in language, mathematics, and reasoning, while Grok-4 Code will serve as a personal coding assistant for top developers [2] - The integration of Grok with Tesla's AI supercomputing system could lead to a transformative leap in AI capabilities, enabling advanced real-time decision-making and multi-tasking in autonomous driving and robotics [4] Group 1 - The release of Grok-4 is scheduled for July 9, with a live stream on xAI's social media platform [1] - Grok-3, released earlier, demonstrated exceptional reasoning and understanding capabilities, outperforming competitors like Gemini and GPT-4 in various benchmarks [1][3] - The AI application software market is experiencing strong demand, as evidenced by robust performance from companies like C3.ai, APPlovin, and Palantir [3] Group 2 - The upcoming Grok models are set to compete fiercely with OpenAI's GPT series and Google's Gemini 2.5 Pro [2] - The integration of Grok into Tesla's AI systems is expected to enhance the performance of both FSD and the Optimus robot, allowing for rapid processing and decision-making [4] - The AI supercomputing system's capabilities, combined with Grok, could create a highly efficient autonomous driving system capable of complex task execution [4]
刚刚,Grok4跑分曝光:「人类最后考试」拿下45%,是Gemini 2.5两倍,但网友不信
机器之心· 2025-07-05 02:46
Core Viewpoint - The leaked benchmark results for Grok 4 and Grok 4 Code indicate significant performance improvements, suggesting that the models may surpass competitors in various AI assessments [2][3][26]. Benchmark Results - Grok 4 achieved a standard score of 35% on the Humanities Last Exam (HLE), which improved to 45% with reasoning techniques, outperforming OpenAI's o3 by two times and GPT-4o by four to five times [3][5]. - In GPQA (Graduate-level Physics and Astronomy questions), Grok 4 scored 87-88%, comparable to OpenAI's top performance and exceeding Claude 4 Opus's score of approximately 75% [6]. - Grok 4 scored 95% on the AIME '25 (2025 American Mathematics Olympiad), significantly higher than Claude 4 Opus's 34% and slightly better than OpenAI's o3, which scored between 80-90% depending on reasoning mode [7]. - Grok 4 Code scored 72-75% on SWEBench, matching Claude Opus 4 and slightly surpassing OpenAI's o3 at 71.7% [8]. Model Development and Features - Grok 4 is designed as a generalist model with capabilities in natural language, mathematics, and reasoning, and it completed training on June 29 [17]. - The model supports approximately 130,000 tokens in context, indicating a focus on optimizing reasoning speed rather than maximizing long-context performance [16]. - Grok 4 Code is tailored for programming tasks, allowing users to ask coding questions directly [18]. Development Process - Elon Musk has been heavily involved in the development of Grok 4, reportedly working overnight to ensure the model's progress, which he described as going well but still requiring final large-scale training [20][23]. - The recent benchmark scores have generated excitement and speculation about the potential release of Grok 4, with expectations that it may be officially announced soon [25][26].
马斯克被曝在公司搭帐篷!Grok 4 提前泄露,号称要重写人类知识库
Sou Hu Cai Jing· 2025-07-04 15:00
Core Insights - The upcoming release of Grok models, including flagship Grok 4 and programming-focused Grok 4 Code, is set for July 4 [2] - Grok 4 Code is designed for programmers, integrating deeply with the Cursor editor for tasks like code completion, bug analysis, and code generation [5] - Grok 4 supports multimodal input, allowing for both text and image processing, enhancing its applicability in various scenarios [6] Technical Features - Grok 4 features a context window of 130k tokens, which is smaller than some competitors but aims for speed and accuracy [6] - The model exhibits strong reasoning capabilities and structured output, making it suitable for diverse applications beyond natural language processing [6] - Grok 4 introduces a native code editor similar to VS Code, with potential "agentic coding" abilities for proactive suggestions and debugging [7] Strategic Goals - Elon Musk has ambitious plans for Grok, aiming to "rewrite the entire online knowledge base" by eliminating errors and misinformation from training data [7] - The release of both models signifies a move towards a comprehensive AI tool that caters to both general users and developers [10] Market Reception - Previous iterations, like Grok 3, faced criticism for not meeting high expectations, leading to skepticism about the new models [9] - Concerns have been raised regarding potential biases or information manipulation in Musk's plan to purify the knowledge base [9]