xAI’s Grok 4.7 Debuts with Enhanced Parameters and SpaceX Data
- Grok 4.7 was released by xAI after experiencing at least five delays since late July.
- The model operates on a substantial scale with 2.1 trillion parameters, a significant increase from Grok 4.6’s parameters.
- Supplemental training data from SpaceX, including Starlink telemetry and engineering logs, enhances its reasoning about hardware systems.
- Grok scored second to Claude Fable on GDPval and AA-Briefcase benchmarks, and second to GPT-6 Astra on EEBench.
- Despite trailing competitors in raw capability, Grok is priced competitively at $2 per million input tokens and $6 per million output tokens.
Grok’s release marks a notable enhancement over previous versions, driven by an increase in parameters and integration of specialized SpaceX data for improved reasoning capabilities in engineering contexts. The model’s competitive pricing strategy aims to make it accessible despite not leading benchmark scores against rivals like Claude Fable and GPT-6 Astra.
While Grok lags behind top models in certain benchmarks, its cost-effectiveness could appeal to users prioritizing affordability over peak performance metrics (Source).
