[Elon Musk: Significant Performance Gap Between Grok V8 and V9] Elon Musk posted on the X platform stating that the recently completed training run for Grok V9 (1.5T parameters) has performed 'exceptionally well,' and this result has not yet incorporated the supplementary training portion of the Cursor data. The foundational model version currently under internal development is V9, which shows significant improvements over V8 in areas such as data cleaning, training methods, and model scale, and has been optimized for the Blackwell architecture to enhance computational efficiency. Musk emphasized that, by comparison, the current external version, v4.2, is built on the V8 foundational model, with a parameter scale of approximately 0.5T, running on the Hopper architecture. It still has certain limitations in terms of training data quality and coverage. The performance gap between Grok V8 and V9 is substantial, with the next-generation model achieving a leap forward in overall capabilities.