Grok 4.5 ranked first on Artificial Analysis’s AutomationBench-AA, posting a 51.4% score and a $0.34 cost per task. According to NS3.AI, Grok 4.5 outperformed Claude Fable 5 and Claude Opus 4.8 on the benchmark.
The results also showed Grok 4.5 recorded more guardrail violations than its closest competitors, despite leading on overall score and cost per task.
AI TRENDS | Grok 4.5 Tops AutomationBench-AA With 51.4% Score but Logs More Guardrail Violations
2026-07-09 07:36:09
Disclaimer:
1. The information provided does not constitute investment advice. Investors should make independent decisions and bear all risks themselves.
2. The copyright of this content belongs to the original author. The views expressed herein are solely those of the author and do not represent the stance or position of this website.
Previous article：
Moqi Intelligence Raises Over CNY 1 Billion in Angel Round Series Led by Alibaba and TencentNext article：
AI趋势 | 华尔街大行缺席香港58亿美元AI融资
