Woofun AI data shows that Vals AI released Muse Spark 1.1 scores, placing the model first in Medical Records, Tax Q&A, and Legal Agent assessments with 88.89%, 79.72%, and 20.00% respectively. In TaxEval v2, it surpassed Fable 5, Opus 4.8, and GPT-5.5 on over 1200 questions, where top ten models differed by less than 3 percentage points.

Muse Spark 1.1 achieved 20.00% in Legal Agent tasks involving real legal work, exceeding Grok 4.5 at 12.92% and Fable 5 at 11.25%. It scored 72.16% in Web Application Programming, beating GPT-5.5 and Grok 4.5 but trailing Fable 5 and certain Claude models.