Grok 4.7 Scores 46 on AI Intelligence Index, Puts SpaceXAI in Top 4 Labs (artificialanalysis.ai)

Grok 4.7 has achieved a score of 46 on the Artificial Analysis Intelligence Index, positioning SpaceXAI among the top four AI labs globally. This latest version, which improves by 2 points from Grok 4.6, demonstrates significant advancements in agentic knowledge work, notably scoring 1657 Elo on the AA-Briefcase benchmark and overtaking previous models like GPT-5.6. With a focus on high reasoning effort tasks, Grok 4.7's analytical quality particularly shines, scoring 1994 Elo, which indicates its enhanced capability in generating practical work outputs. The model's notable leap in coding performance is also attributed to Grok Build, with an increase of 9 points on the Artificial Analysis Coding Agent Index, now ranking fourth overall. However, these advancements come at a cost—increased token usage has surged to approximately 81,000 tokens per task, significantly more than its predecessor Grok 4.6. While Grok 4.7 maintains a similar context window and pricing structure, its enhanced capabilities and the strategic use of tokens highlight a critical evolution in AI model design, suggesting implications for the future of agentic AI solutions in professional environments.