The promised release of Grok 3 by xAI, Elon Musk’s company, did not happen on time. Last summer, Musk promised that this model, which is meant to rival GPT-4 from OpenAI and Gemini from Google, would appear by the end of 2024. It was supposed to be trained on a massive GPU cluster in Memphis, which was expected to ensure a significant breakthrough in AI capabilities. However, at the beginning of January, Grok 3 still hadn’t appeared, and there are no signs of its imminent release.
Interestingly, there have been hints about an intermediate model, Grok 2.5, on the xAI website. Well-known insider Tibor Blaho discovered code indicating a possible release of this model, which could become the company’s “smartest” development to date.
This is not the first time Musk has missed his ambitious deadlines. In one interview, he even admitted that Grok 3 might only appear “if we’re lucky” in 2024.
This delay is part of a broader trend in the AI world. Last year, Anthropic failed to release the planned Claude 3.5 Opus model, and both Google and OpenAI also faced issues with their flagship developments. This likely points to the limitations of current AI scaling methods, where increasing computing power and data volumes no longer yield the expected results.
