On July 8, Elon Musk, CEO of Tesla (TSLA), said that xAI, the artificial intelligence startup he founded and helmed, plans to release its latest version of AI big model Grok-4 on July 9 (Wednesday), Pacific time.
Grok-4 big model is coming soon
For Tesla FSD (full automatic driving system) and the “Optimus” humanoid robot (i.e. Optimus), which Musk highly praised, it is expected to usher in a “super upgrade” based on the reasoning artificial intelligence big model.
In February of this year, the emergence of the Grok-3 artificial intelligence big model triggered a “big model reasoning storm” – Grok-3 showed extremely powerful reasoning ability, logical adaptability, and understanding of extremely complex physical knowledge at the press conference. At that time, it defeated all big model competitors including Gemini, Claude, and GPT-4o in mathematics, science and programming benchmarks.
According to insiders, xAI is about to launch Grok 4 and Grok 4 Code. Among them, the flagship model Grok 4 is expected to perform better than OpenAI and Google’s latest AI models in terms of language, mathematics and reasoning; Grok 4 Code is positioned as a personal coding assistant for the world’s top developers, and can even be embedded in the user’s customized code editor.
Apple acquires two AI companies
At the same time, it is reported that Apple (AAPL) recently quietly completed the acquisition of two technology companies. This move was interpreted by the outside world as an important move to accelerate its technology layout in the fields of Apple Vision Pro headsets and Apple Intelligence artificial intelligence.
According to MacGeneration, Apple’s attention to artificial intelligence-related companies has increased significantly, and it has brought at least two AI companies under its command. One of them is TrueMeeting, whose core technology allows users to scan their faces through iPhones to generate AI avatars.
Almost at the same time as the acquisition of TrueMeeting, Apple also acquired WhyLabs. WhyLabs’ core business is to monitor large-scale language models of artificial intelligence, and its key role is to prevent the model from “hallucination”.
In addition, facing rivals such as OpenAI and Google, Zuckerberg has made the development of AI technology the top priority for Meta (META). He personally participated in recruiting the top talents in the field of AI development, entertained potential AI candidates in his mansions in Silicon Valley and Lake Tahoe many times, and personally contacted talents.
Meta management previously stated that it will invest tens of billions of dollars in AI-related fields this year, mainly for the construction and expansion of data centers, and spend a huge amount of money to purchase AI computing power infrastructure such as AI chips.
Industry insiders pointed out that with the current development of AI technology, with DeepSeek’s comprehensive leadership of the “extremely low-cost AI training/inference” new paradigm, it has greatly streamlined the full range of AI workloads from AI large model training to deployment – AI training costs have dropped significantly and the token cost on the inference side has dropped sharply, covering all walks of life on the B-side or C-side in the future. Killer generative AI applications.
WiMiAI layout reconstructs high-end battlefield
According to the data, Wimi Hologram Cloud Inc (WIMI), as an innovative participant in the field of artificial intelligence, focuses on “high-end battlefields” such as high-performance computing power, multimodal models, embodied intelligence, quantum fusion, etc., and accelerates the reconstruction of the core competitiveness of the AI industry through the construction of computing power base, model algorithm iteration, deep penetration of scenarios and ecological synergy, and accelerates the upgrading of technology to “high value and strong barriers”.
At present, computing power is the core driving force of AI, and WiMi focuses on “autonomous and controllable computing power” as the layout focus, supporting large model training, reasoning and complex scenario applications through high-performance computing power bases and customized chips. At the same time, WiMi keeps up with the trend of open source large models, develops AI applications such as natural language processing (NLP) and code completion based on DeepSeek, reduces application costs through algorithm optimization, and promotes the large-scale penetration of generative AI in scenarios such as e-commerce live broadcast and digital marketing.
Conclusion
As global companies compete for the high ground of AI applications, AI visual reasoning drives large models to move one step closer to general multimodal reasoning. Large model companies are also seeking to transform this capability into digital productivity and seek commercialization. Therefore, the track of AI large models is infinitely broad. Language and multimodal models, AI can be used for innovation and research and development, which may greatly accelerate the speed of human innovation and will continue to attract high attention.
