Musk: Grok-3 Will Be “The Most Powerful Model on Earth”
“One hundred percent, people are going to love Grok!” Elon Musk stated during a live broadcast on February 18, Taiwan time, for the launch of the xAI model he founded.
Due to Musk’s announcement the previous day on social media that the upcoming Grok-3 would be “the most powerful model on Earth,” the official live broadcast attracted over two million simultaneous viewers. Grok-3 is compared to Google’s Gemini, OpenAI’s ChatGPT, Anthropic’s Claude, and also includes China’s startup “DeepSeek” in the comparison. During the broadcast, the xAI team demonstrated Grok-3 and its smaller version, Grok-3 mini, showing that in tests of mathematics, science, and computer engineering, its performance exceeded that of other existing models.
Grok-3 outperforms benchmark models in mathematics, science, and programming.
Image / xAI
On the model blind testing platform Chatbot Arena (formerly LMSYS), the early Grok-3 model, codenamed Chocolate, received a user rating of 1,400 for comprehensive Q&A, outperforming Gemini 2.0 and ChatGPT-4o. Musk emphasized that Grok-3’s computational capability is over ten times that of the previous generation, and it will continue to adjust dynamically once online, stating, “This model will improve every day!”
Grok-3 Introduces Three New Modes for In-Depth Research, Thinking, and Creative Applications
DeepSearch Mode: Specifically for Solving Complex Problems
One of xAI’s goals is to solve complex problems, including Musk’s focus on space travel and other scientific issues that require extensive data analysis and complex calculations. Following Google, OpenAI, and the AI search engine Perplexity introducing features aimed at scientific research like “Deep Research,” xAI has also added the “DeepSearch” mode to Grok-3, which can conduct comprehensive research, thinking, and presentation of analytical results for complex issues. Musk stated that “DeepSearch” can be seen as “the next generation of search engines,” allowing AI to complete research that previously took hours within ten minutes. Notably, users can select the “Show Thinking” feature to see the complete process of how the AI understands and processes problems, changing the passive nature of content consumption by users and making the content traceable and transparent.
Grok-3 includes the deep research mode “DeepSearch.”
Image / xAI
Think Mode: Proficient in Reasoning and Physics Problems
In addition, the Grok-3 model also includes “Think” and “Big Brain” modes. During the live broadcast, xAI engineers indicated that the “Think” mode excels at handling high-level reasoning and physics problems, such as having Grok-3 write a program to calculate celestial movements.
Grok-3’s Think mode can also adjust thinking speed.
Image / xAI
Big Brain Mode: Skilled in Abstract Creation
As for the third mode, “Big Brain,” Musk referred to it as “the starting point for AI creativity,” capable of engaging in relatively abstract creation. For example, the team demonstrated during the broadcast how Grok-3 could generate a new game combining Tetris and color block elimination.
Grok-3’s “Big Brain” mode can leverage creativity.
Image / xAI
The xAI team also stated that they will quickly develop a voice-based chatbot for Grok-3. Currently, Grok-3 has been released to X’s Premium+ subscribers, and xAI is working on a new subscription plan called “SuperGrok” for the app and website, covering Grok-3 along with the “DeepSearch” and “Think” features.
Image / xAI
Musk indicated that Grok-3’s API is expected to be made public in a few weeks, and as per tradition, xAI will open-source the previous generation model once the latest model stabilizes, with Grok-2 expected to be open-sourced in the coming months.
This article is authorized for reprint from: Digital Age