What trump card has Elon Musk prepared by releasing a ChatGPT competitor?

Elon Musk open-sourced xAI's Grok model, a chatbot similar to ChatGPT. Let's take a closer look at what this means for the development of AI models.

Сreating open source AI models became very common. For example, the Bloom or Falcon LLMs were originally created by non-profit organizations as open source models aimed at democratizing AI technologies. Сommercial companies have also followed this path. For example, Meta decided to share its Llama-2 artificial intelligence as open source. "Our open source large language model is now free and available for research and commercial use. This release offers a unique opportunity for developers, while reflecting our commitment to open source, cross collaboration and innovation," explained Meta.

In some of his posts on the social network he owns, X, Elon Musk has recently criticized the fact that, contrary to its name, OpenAI does not make its technologies open source. He has also moved from words to action. In 2023, his xAI Artificial Intelligence Lab introduced a model similar to ChatGPT called Grok. On March 11, 2024, Musk announced that Grok will be an open source model. "This week xAI will make Grok open source," he wrote in a post on the social network he owns, X (formerly Twitter). Tesla boss didn't give any further explanation at the time, but many assumed the decision had something to do with the conflict that currently exists between the Musk and OpenAI, a company he actually co-founded.

At first, the Grok chatbot was only available to Premium+ paid subscription program users, and for that reason, it was not used by as many people and companies compared to models like Meta's Llama-2 or Google's Gemini. So what does the open sourcing of the new bot give us and what the Grok model can do?

The open release of the chatbot includes the basic model weights and the Grok-1 model network architecture with 314 billion expert mixture parameters. The current version of Grok is provided under the terms of the Apache 2.0 license, which means you can use it commercially, but that license does not include the data used to train it, or the connection to X for real-time data. xAI Artificial Intelligence Lab's blog states that it is intended for use in coding, creative writing, and question answering but has not been fine-tuned "for any specific application, such as dialog". In other words, Musk's LLM still needs to undergo further training and improvement in order to compete on an equal footing with leading AI models such as Google's Gemini or OpenAI's ChatGPT. However, in the digital world everything develops very rapidly.