Grok 4.7 Tops New Legal AI Benchmark With 9.4% Score

xAI's Grok 4.7 leads the Harvey LAB-AA v1.1 legal AI benchmark with a 9.4% Hallucination-Gated All-Pass Rate, followed by Meta's Muse Spark 1.3 and OpenAI's GPT-6 Astra. The benchmark penalizes models for fabricated content or contradictions, and Grok 4.7 costs approximately $9.50 per task.