Grok‑3 is the latest large language model (LLM) from Elon Musk’s xAI, released on February 17, 2025. It’s trained with a massive 10× more compute than Grok‑2 and outperforms GPT‑4‑level benchmarks in math reasoning and science.