xAI Unleashes Its Secret Weapon: Grok V9-Medium with 1.5 Trillion Parameters Completes Training, Targeting the AI Programming Arena

The global AI computing race is intensifying again. On May 25, Elon Musk officially announced that xAI's latest flagship model, Grok V9-Medium, has successfully completed the training phase. As a massive foundational model with 1.5 trillion (1.5T) parameters, its scale is exactly three times that of the current v8-small version (0.5T parameters) that supports all Grok production traffic.

Key Highlights:
Scale Leap: From 0.5T to 1.5T parameters, the model has achieved a qualitative leap in reasoning depth and complex task handling capabilities.
Programming Specialization: During the supplemental training phase, xAI incorporated a vast amount of Cursor (AI programming tool) code data to significantly enhance its ability to handle complex programming tasks.
Timeline: The model is currently in the supervised fine-tuning (SFT) phase, and reinforcement learning (RL) will start soon. It is expected to be officially released to the public within 2 to 3 weeks.
Architecture Optimization: Musk revealed that the model has been deeply optimized for NVIDIA Blackwell architecture GPUs, which will significantly improve computing efficiency.

New Landscape of "Programming AI": xAI Faces Off Directly with Top Code Assistants

The release of Grok V9-Medium is particularly notable for its "programming DNA." Musk previously admitted that the existing v8-small version had obvious shortcomings in the quality, comprehensiveness, and balance of its training data, and V9-Medium is a systematic reconstruction targeting these issues.

By incorporating Cursor's code logic and practical data, xAI aims to establish a significant technical advantage within the developer ecosystem. For developers, Grok V9-Medium is not just a general chatbot but could become an "AI engineer" capable of deeply understanding complex codebases and independently completing complex programming logic.

Why Add Cursor Data?

As a mainstream AI programming assistant, Cursor represents the code calling habits, engineering thinking, and bug fixing paths of top software engineering practices. By integrating this data into "supplemental training," xAI means that Grok will skip the stage of a simple language model and directly enter the "engineering practice" stage.

This approach is seen by industry observers as xAI's "overtaking on a different track": instead of only training through general corpora, it directly learns the most authentic and direct thinking patterns of human professional engineers in programming tools.

Market Expectations: The "Benchmark" Battle of Large Models Will Be Rewritten

With Grok V9-Medium expected to launch in mid-June, competition in the AI field will enter a new phase of intense rivalry:

Rebuilding Performance Standards: The 1.5T parameter scale means it will directly compete with the most advanced multimodal models in logical reasoning and complex instruction following.
Benefits of the Blackwell Architecture: As one of the first AI large models deeply adapted to the Blackwell architecture, its performance in processing efficiency and inference cost will directly affect xAI's deployment strategy on Tesla vehicle edge devices and X platform applications.

Google CEO Sundar Pichai: Frankly Admits Lags in Programming Field, AI Search Transition Will Take a Steady Approach

After the I/O conference, Google CEO Sundar Pichai accepted a podcast interview, openly discussing the AI competition. He admitted that Google leads in overall intelligence and multimodal capabilities, but lags in agent-based programming and complex long-term tasks. The conversation also touched on changes in search patterns and AI safety, reflecting Google's open attitude as it faces challenges in the AI race.

Powerful Collaboration! Texas Plans to Build a 2.5 Gigawatt Nuclear and Gas Hybrid Power Plant to Address the Surge in AI Energy Demand

U.S. energy company Blue Energy has partnered with GE Vernova to develop a 2.5 gigawatt hybrid power plant in Texas, incorporating both nuclear and natural gas power facilities to address the increasing electricity demand from AI data centers. The project pioneers the "gas to nuclear" model, using natural gas for quick power supply and gradually transitioning to nuclear energy, effectively addressing the challenges of long approval and construction cycles for traditional nuclear power plants. It provides a flexible and stable power solution for high-energy-consuming industries.

Zhangzhou Trials a New Model of Solo Entrepreneurship: AI Empowers Individuals to Compete in Digital Trade

In the AI era, the form of entrepreneurship is undergoing a 'streamlining' revolution. On March 12th, Longwen District in Zhangzhou, Fujian Province, launched the AI+OPC Entrepreneurship Investment Conference and the first AI+OPC Entrepreneurship Challenge, and unveiled the province's first AI+OPC Promotion Development Center. This marks the official industrialization of the entrepreneurial model centered on 'solo teams' driven by artificial intelligence. OPC stands for 'Solo Company,' where individuals can independently manage business operations using AI tools such as large models and intelligent agents, achieving efficient entrepreneurship.

Microsoft Launches New Fara1.5 Series AI Agent Model with a Task Success Rate of 72%! Exceeding OpenAI!

Microsoft Research has launched the Fara1.5 series of AI agent models, specifically designed for browser scenarios, including 4B, 9B, and 27B parameter versions. The model works with MagneticLite sandbox browser to directly read screenshots and simulate mouse and keyboard operations, automatically completing web tasks and enhancing the intelligence level of computer usage.

Q1 AI Investment Booms! Domestic Large Models See Funding Surges to 30 Billion Yuan!

In the first quarter of this year, there were nearly 600 AI-related financing events with a total amount exceeding 110 billion yuan, a year-on-year increase of 185.4%. Domestic large model projects such as "Moonlight Darkness" and "Step Star" secured over 30 billion yuan in funding in May, and the industry trend continues to heat up.