xAI announced its Grok 4.7 model, developed for complex tasks focused on coding and information. The model has been improved in long-horizon problem-solving, output verification, and safety, and is offered in the same price and speed range as Grok 4.6.
Grok 4.7 delivered strong price-performance results in CursorBench 4.0 tests while outperforming Grok 4.6 in the GDPval and AA Briefcase tests. The model is now available through Cursor, Grok Build, Grok API, third-party coding tools, and cloud platforms.
Input tokens are priced at 2 dollars per million, while output tokens are priced at 6 dollars per million. A version that generates output twice as fast is offered at twice the price.
Why it matters
This announcement broadens the evaluation framework for developers and organizations choosing models, particularly for coding and information-intensive work. The model’s availability across different tools and cloud platforms allows users to compare performance and costs without completely changing their existing workflows. The division of pricing into input, output, and faster-use options means that, for teams with high production volumes, not only technical capacity but also usage patterns will be decisive in the decision-making process. However, the announced benchmark results demonstrate performance in specific tests; how the model will perform in real-world workflows across different industries, particularly in tasks requiring long-horizon problem-solving and output verification, remains an open question.
Background
Grok is not a new name in the FikirPilot archive: over the past 90 days, we published 2 articles mentioning the name, the latest dated September 6, 2026.