I was really looking forward to Grok 4.7, but it seems it was all hype.
What's funny is that Grok 4.6 was such a major leap in such a short time. It was, and still is, good all around, both in real-world use and on benchmarks. Price, token efficiency, capabilities: all basically exactly what was needed.
It seems the xAI team has hit a wall and is now facing diminishing returns in their model training. Hopefully they've recognized this and corrected their approach for future model development.
grok 4.7's frontend capabilities is rough... to say this least
this is with the frontend design skill btw