Musk: Grok Foundation Model V9 nears release after training completion
Elon Musk has announced that xAI’s latest large language model, Grok Foundation Model V9-Medium, has completed its pre-training phase and is now moving into final stages ahead of public release.
Posting on X, Musk said the 1.5 trillion-parameter model has finished training, with internal evaluations described as strong. He added that additional training data from Cursor has been incorporated, with more supplementary data expected in the next phase.
The development process is now entering fine-tuning, while reinforcement learning is set to begin within days. A public release is projected in two to three weeks, depending on final optimisation results.
Musk said the new model represents a substantial improvement over the current Grok v8-small system, which uses a 0.5 trillion-parameter architecture and currently powers production traffic. He highlighted especially stronger performance on complex coding and reasoning tasks.
By Aghakazim Guliyev







