**[Meta Gains 11 Points in Four Months, Muse Spark 1.2 Rises to Third in the U.S.]** According to monitoring by Beating Insights, after Meta released Muse Spark 1.2, the evaluation agency Artificial Analysis gave it a score of 54, which is 3 points higher than the previous version and 11 points higher than the initial version released in April this year. It is now tied with Grok 4.5 and just 1 point behind GPT-5.5, ranking third among AI models in the U.S. This improvement is primarily attributed to advancements in Agent tasks. In the real-world knowledge work evaluation GDPval-AA v2, Muse Spark 1.2's Elo rating increased from 1371 to 1631, climbing to fifth place overall, surpassing Claude Opus 4.8. Terminal operations and banking tool integrations also saw improvements, while performance in scientific reasoning remained largely unchanged. The pricing has not increased, with costs per million input and output tokens still at $1.25 and $4.25, respectively. Artificial Analysis estimates that completing a comprehensive evaluation task costs an average of $0.40. In comparison, GPT-5.5 costs $1.18, and Kimi K3 costs $0.86. [Original Link]