Article

Musk: Grok 4.8 has completed pre-training, small-scale rollout to begin later this week

PANews November 10 news, Musk posted disclosing the progress of Grok 4.8: pre-training is complete, currently in the reinforcement learning (RL) stage, evaluation scores are starting to rise. xAI plans to first push a small-scale rollout (soft rollout) to Bot users later this week. Musk stated that the team has abandoned JAX/SGLang and rewritten the entire reinforcement learning and inference system from scratch using C/C++, which is one reason for the delay in progress; supervised fine-tuning (SFT) data has also undergone extensive cleaning. He described Grok 4.8 as "completely different" from 4.7, saying it is "not yet outstanding," but the underlying foundation has been built.

Source: https://www.panewslab.com/zh/articles/01a1c4fe-0dde-7074-be44-21baa2d23bb3