Meta's Muse Spark 1.1 model now holds the State-of-the-Art (SOTA) position on Harvey's Legal Agent Bench, TaxEval, and MedScribe benchmarks. The model outperforms Fable 5 on MedScribe and TaxEval, while being 10 times cheaper and twice as fast. Muse Spark 1.1 also dethroned Grok 4.5 on Harvey's Legal Agent Bench.