xAI prepares Grok 4 and Grok 4 Code for upcoming launch

xAI appears to be in the final stages of rolling out its next-generation model, Grok 4, with updated references surfacing in the xAI developer console. Two model variants, Grok 4 and Grok 4 Code, are now listed, along with a banner that suggests public availability is imminent. According to the banner, Grok 4 will be accessible via API and will support text and vision inputs, with image generation and other modalities still in development. This aligns with Elon Musk’s earlier statement that Grok 4 would launch on July 4, now just days away, signalling that deployment preparations are actively underway.

The new model card hints at a 130k token context window, smaller than many competing frontier models, which may suggest xAI is optimizing for inference speed and real-time usability rather than maximal long-context performance. Feature-wise, Grok 4 will include function calling, structured outputs, and reasoning capabilities. It’s being positioned as a generalist model with strengths in natural language, math, and code understanding.

Grok 4 Code, on the other hand, appears to be a coding-optimized variant and is described as a companion model for code-related tasks. The model will be tightly integrated with tools like Cursor, a code editor where Grok Code will likely be available on launch day. The target audience includes developers looking for embedded assistance within their coding environments.

This dual-model release reflects xAI’s intent to compete directly in both the general-purpose AI space and the developer-focused assistant segment. Unlike earlier versions, Grok 4 seems to be intended for broader deployment across API and web interfaces simultaneously. If it sticks to the July 4 timeline, Grok 4 could mark a step-change in how xAI scales its assistant across consumer and developer platforms.