Elon Musk, founder of xAI, has unveiled the company’s first AI product, a Large Language Model (LLM) called “Grok.” Musk announced the release of Grok on his X social network, stating that it is currently available to a limited number of users in the United States. Users can join a waitlist to gain early access to Grok. The product is designed to compete with other leaders in the space, such as OpenAI’s GPT and Anthropic’s Claude 2. Grok is described as an AI modeled after the Hitchhiker’s Guide to the Galaxy, with a sense of humor and the ability to provide real-time information.
According to the xAI website, Grok was trained on 33 billion parameters of data, compared to 70 billion for Meta LLama 2 and 20 billion for OpenAI’s GPT-3.5 models. xAI claims that Grok-0 approaches the capabilities of Meta LLama 2 while using only half of its training resources. The xAI team has made improvements in reasoning and coding capabilities, creating a new model called Grok-1, which powers the Grok chatbot client.
Grok’s performance in machine learning benchmarks surpasses other models in its class, including ChatGPT-3.5 and Inflection-1, according to xAI. The company has showcased the rapid progress it is making in training LLMs with exceptional efficiency. Grok is intended to answer questions and suggest what questions to ask, with a sense of wit and humor.
Elon Musk shared screenshots of Grok’s responses to user prompts, demonstrating its humor and sarcasm. Grok has real-time access to information through the X platform (formerly Twitter), providing a unique advantage over other models. Musk emphasized that Grok’s current knowledge is derived from the X platform, though it is not disclosed how xAI is using the platform or users’ posts to train Grok.
Musk also announced that Grok’s availability will be expanded to all X Premium+ subscribers once it is out of early beta. This move indicates Musk’s intent to compete with OpenAI, as the company prepares to announce new AI features at its upcoming developer conference.
VentureBeat aims to provide a platform for technical decision-makers to gain knowledge about transformative enterprise technology.
