Competition in the artificial intelligence market is reaching a new fever pitch. In just ten days, on September 12, xAI will unveil the long-awaited Grok 4.7 model. Elon Musk, in his public statements, is not merely announcing an update but making an extremely bold prediction: the new version will surpass all existing AI models to date. This statement sounds particularly ambitious against the backdrop of a recent announcement from a competitor — OpenAI's Astra model.
Secret weapon: SpaceX engineering data
The key difference between Grok 4.7 and its predecessors lies not only in scale but also in data. Musk has confirmed that the model's base training is complete, and the final stage is now underway, during which internal developments and SpaceX telemetry are being integrated into the architecture. This is an unprecedented step that could give the model a unique advantage in solving complex engineering and scientific problems that require real-world knowledge, not just information from open internet sources.
The technical specifications are also impressive. Grok 4.7 will rely on 2.1 trillion parameters, significantly more than the 1.5 trillion in Grok 4.6. According to Musk, the increase in parameters will lead to a slight slowdown in performance, but this is offset by more economical token consumption and, consequently, high efficiency.
Benchmarks and competition
It is worth recalling that the previous version, Grok 4.6, released on August 12, already set a high bar. In the AA Intelligence index, it scored 61 points, tying with GPT-5.6 Sol Max but falling one point short of the leader — Claude Fable 5 Max (62 points). However, in the Terminal-Bench v3.0 test, the model achieved only 26%, significantly trailing OpenAI's competitor with its 34.6%. This indicates that xAI has areas for growth, particularly in the field of autonomous AI agents.
Musk himself does not hide that he is closely watching competitors and calls Anthropic a "great company," expecting more advanced solutions from them soon. Nevertheless, he is confident in his superiority, betting on unique SpaceX data that other market players are unlikely to obtain.
My analysis: The integration of proprietary SpaceX data is indeed a powerful move that competitors will find difficult to replicate. However, as practice shows, loud promises and real results on independent benchmarks do not always align. The key moment will be xAI's publication of third-party test results after the release. If Grok 4.7 truly demonstrates a breakthrough in engineering disciplines, it could radically change the balance of power in the AI market and strengthen Musk's influence in the technology industry.