Recently, the next-generation AI model Grok3, originally scheduled for release by Elon Musk's AI company xAI by the end of 2024, has not materialized as expected, drawing significant attention from the industry. Grok3 is seen as xAI's major competitive product against OpenAI's GPT-4 and Google's Gemini, featuring image analysis and question-answering capabilities, and has already been applied on Musk's social network X.
Last summer, Musk stated on the X platform that Grok3 would be “a significant breakthrough after 100,000 H100 training sessions.” However, on January 2, just after the New Year, Grok3 still had not been released, and there were no signs indicating its imminent launch. Reports suggest that an intermediate version, “Grok2.5,” may be released first.
This situation is not an isolated incident; several companies in the AI industry have experienced similar awkward circumstances. For instance, AI startup Anthropic failed to launch its successor to the Claude3Opus model, Claude3.5Opus, on schedule last year and ultimately decided to abandon the release of the model after announcing it would be launched by the end of 2024. Additionally, both Google and OpenAI have faced delays in releasing their flagship models.
The series of delays may reflect the bottlenecks currently facing AI training methods. In the past, companies could achieve significant performance improvements through large-scale computing power and larger datasets. However, as the performance gains of each model generation have gradually diminished, companies are beginning to seek alternative technologies to break through the current bottlenecks. Musk acknowledged in an interview with podcast host Lex Fridman that while expectations for Grok3 are high, he also admitted it might not be realized as hoped.
Moreover, the relatively small size of the xAI team may be one of the significant reasons for the delay of Grok3. Compared to its competitors, xAI has limited resources and personnel, undoubtedly posing more challenges in developing new models.
Overall, the delay of Grok3 is not just a minor episode for xAI but reflects a broader trend within the entire AI industry when faced with technological bottlenecks.
