SpaceXAI unveils Grok 4.7 with upgraded work capabilities

SpaceXAI has released Grok 4.7, an updated artificial intelligence model tailored for software engineering and extended workplace tasks. According to the company's official announcement, the new version features enhanced self-verification routines and handles complex projects that require hours of sustained reasoning.

The model maintains the exact pricing structure of its predecessor, setting costs at two dollars per million input tokens and six dollars per million output tokens. SpaceXAI reported that Grok 4.7 outperforms older versions on several industry benchmarks for programming, legal review, and general office work without increasing latency.

Alongside raw performance upgrades, the developer introduced a redesigned safety system aimed at blocking malicious cyber activities while allowing legitimate security research. Access to Grok 4.7 has opened immediately across development platforms like Cursor, the Grok API, and custom cloud integrations.

While keeping pricing stable during model upgrades is a welcome trend for developers, it remains unclear whether these incremental reasoning benchmarks translate to real-world productivity gains for non-technical office workers. Moreover, as companies build increasingly autonomous agents for multi-hour projects, enterprise customers will likely demand stronger guarantees around reliable output quality over extended run times.