Meta: Muse Spark 1.1

Meta's Muse Spark 1.1, a multimodal reasoning model designed for agentic tasks, demonstrates moderate speed performance, ranking in the 31st percentile across five benchmarks. Its pricing tends to be at premium levels, placing it in the 14th percentile. A significant strength is its exceptional reliability, achieving a 100% success rate across all benchmarks, indicating consistent and usable responses without technical failures. In terms of specific performance, Muse Spark 1.1 excels in Instruction Following, achieving 86.2% accuracy and ranking in the 92nd percentile, showcasing strong precision in complex directives. Its Coding performance is solid at 86.9% accuracy (50th percentile). However, it shows a notable weakness in Email Classification, with an accuracy of 89.0%, placing it in the 10th percentile. Ethics performance is also relatively low at 84.0% accuracy (18th percentile), suggesting areas for improvement in moral reasoning. Mathematics accuracy is moderate at 87.0% (43rd percentile). Overall, the model's key strengths lie in its robust instruction following and outstanding reliability, while its primary weaknesses are its premium cost and lower performance in classification and ethical reasoning tasks.