[Meta's New Model Secures First Place in Three Rankings, Legal Agent Surpasses Grok 4.5] According to monitoring by Beating, AI evaluation organization Vals AI has released the results for Muse Spark 1.1. The model ranked first in evaluations for medical documentation, tax Q&A, and legal Agent, with scores of 88.89%, 79.72%, and 20.00%, respectively. In the TaxEval v2 tax evaluation, it outperformed Fable 5, Opus 4.8, and GPT-5.5. This evaluation includes over 1,200 unpublished test questions, assessing both answers and reasoning processes. The gap among top models is minimal, with less than a 3-point difference among the top ten. In the legal Agent evaluation, Muse Spark 1.1 scored 20.00%, surpassing Grok 4.5's 12.92% and Fable 5's 11.25%. The test required models to use documents, tables, and file tools to complete real-world legal tasks. In the web application programming evaluation, it scored 72.16%, outperforming GPT-5.5, Grok 4.5, and GPT-5.4, though it still lags behind Fable 5 and several Claude models. [Original Link]