Part 4/8:
Artificial Analysis reported scores that positioned Llama 4 as a formidable contender within the AI intelligence index, especially for non-reasoning models. Maverick scored 49, putting it ahead of Claude 3.7 Sonnet, while Scout managed to stay in line with GPT-4 Mini. The critical takeaway is the efficiency: Llama 4 manages to maintain high performance with significantly fewer active parameters, which translates to lower operational costs—approximately $0.24 per million input tokens compared to the steeper costs associated with its closed-source counterparts.