Model
49d agoMeta Muse Spark 1.1 Beats GLM-5.2 In Coding
A score of 71.3 puts Muse Spark 1.1 ahead of GLM-5.2 in coding benchmarks. Meta reduced the model's hallucination rate from 73 to 38 percent while lowering task costs to $0.26. These efficiency gains improve reliability for developers.
The Signal
The model also climbed eight points on the Artificial Analysis Intelligence Index in three months.