xAI introduced Grok 4.7 for programming and professional tasks

On September 21, xAI unveiled its new artificial intelligence model, Grok 4.7. The company describes it as the most powerful model for programming and professional information processing.

According to xAI, Grok 4.7 handles long and complex tasks more effectively, thoroughly verifies its results, and can work with longer contexts. The model has also been trained to work with Grok Bot tools, which should enhance its capabilities in dialogues and other tasks.

xAI notes improved performance compared to Grok 4.6 in tasks related to document and presentation creation. The model was tested on tasks performed by specialists in various fields, including law, medicine, and financial analysis.

On CursorBench 4.0, Grok 4.7 achieved a score of 46.3%, and on DeepSWE v1.1 — 71%. In EEBench, the model scored 64%, and in Terminal-Bench 4.0 — 37.6%. xAI presents these results in comparison with other models, including GPT-5.6 Sol and Fable 5.1.

The company paid special attention to security. According to xAI, Grok 4.7 features a new system of protective mechanisms and demonstrates higher resistance to attempts to bypass restrictions. In the LatchBio biosecurity test, the model scored 62.4%. In HackerBench v0.3, the share of missed risky requests was 3.3%.

Grok 4.7 is already available in Cursor and Grok Build, as well as through the Grok API, third-party programming tools, model routers, and cloud platforms.

Pricing starts at $2 per million input tokens and $6 per million output tokens. xAI also offers an accelerated version of the model with double generation speed at double the cost.

Source: x.ai