Select Language:
Elon Musk announced today that the Grok foundational model, version V9-Medium, boasting 1.5 trillion parameters, has successfully completed its training phase. This development marks a significant milestone in Musk’s ongoing AI projects. The model’s evaluation results have reportedly been very promising.
During the additional training process, a substantial amount of Cursor data was incorporated to enhance the model’s capabilities. Musk also mentioned that more data will continue to be integrated in the coming weeks to further improve performance.
Focusing on the current advancement, Musk stated that fine-tuning is underway, with reinforcement learning scheduled to commence within the next few days. He estimates that the model will be ready for public release in approximately two to three weeks.
Musk emphasized that this new model represents a major leap forward, particularly in tackling complex programming tasks. Compared to the existing V8-Small version, which handles all Grok production traffic and features 0.5 trillion parameters, the new V9-Medium will be a substantial upgrade.
In related news, Musk’s companies continue to consolidate AI efforts. SpaceX, which acquired the AI startup xAI in February, now oversees xAI’s leadership and operations. Recently, Musk announced that xAI has been renamed SpaceXAI, signaling its integration into SpaceX’s broader AI ecosystem. This move underscores Musk’s commitment to advancing AI technology across his ventures.
ChatGPT
Perplexity
Gemini AI
Grok AI
