Meta: Muse Spark 1.3

Meta's Muse Spark 1.3, a multimodal reasoning model designed for long-running agentic and coding workflows, demonstrates moderate speed performance, ranking in the 35th percentile across benchmarks. Its pricing tends to be premium, positioned at the 10th percentile. A standout feature is its exceptional reliability, achieving a 100% success rate across all evaluated benchmarks, indicating consistent and usable responses. The model exhibits strong performance in several key areas. It achieves perfect accuracy in both General Knowledge and Ethics, notably being the most accurate model at its price point and among models of comparable speed in these categories. Instruction Following and Coding also show high accuracy, ranking in the 96th and 97th percentiles respectively. Reasoning and Mathematics benchmarks also reflect strong capabilities, scoring in the 84th and 81st percentiles. While its Hallucinations accuracy is 96.0% (50th percentile), indicating a reasonable ability to acknowledge uncertainty, its Email Classification accuracy is 98.0% but ranks in the 48th percentile, suggesting this area might be less competitive compared to other models. Overall, Muse Spark 1.3 excels in complex reasoning, ethical understanding, and coding tasks, making it well-suited for its intended agentic and coding applications despite its premium cost and moderate speed.