Grok 4 is xAI’s flagship model released on July 9, 2025, positioned as a reasoning-first successor to Grok 3 with heavier test-time compute and a redesigned training pipeline. At launch, Artificial Analysis measured it at the top of its Intelligence Index, ahead of OpenAI’s o3 and Google’s Gemini 2.5 Pro at the time, and xAI reported an all-time-high GPQA Diamond score near 88 percent along with a joint-best AIME 2024 score around 94 percent. It supports a 256,000-token context window and handles both text and images, and xAI made it available through the Grok app, X, and the API at $3 per million input tokens and $15 per million output tokens. Grok 4’s launch drew attention beyond benchmarks too, including scrutiny over some of its unfiltered responses on X, which xAI addressed with follow-up moderation changes. It set the baseline that xAI’s later Grok 4 Fast and Grok 4.5 releases built on through 2025 and 2026.