Elon Musk Announces Upcoming Releases of Grok 4.6 and Grok 4.7
Elon Musk announced that xAI will release the 1.5-trillion-parameter Grok 4.6 around August 7, followed a few weeks later by the 2.1-trillion-parameter Grok 4.7. These releases follow the recent public launch of Grok 4.5 on July 8. This rapid release schedule highlights xAI's aggressive scaling strategy to compete with top-tier AI models. The massive parameter sizes of Grok 4.6 and 4.7 could set new benchmarks for commercial LLM capabilities. Grok 4.6 is expected to feature significant improvements in Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL). Musk previously described the predecessor Grok 4.5 as an "Opus-level" model with lower inference costs and higher token efficiency.
## BACKGROUND
Supervised Fine-Tuning (SFT) is a process where a pre-trained language model is further trained on a curated dataset of high-quality, labeled examples to improve its performance on specific tasks. "Opus-level" refers to Claude 3 Opus, the most capable model size in Anthropic's Claude 3 family, which serves as a benchmark for high-performance LLMs.