With that said, there are some new features and improvements that are not just about alignment.
CharXiv Reasoning: In "figure understanding," Muse Spark achieved a score of 86.4, significantly outperforming Claude Opus 4.6 (65.3), Gemini 3.1 Pro (80.2), and GPT-5.4 (82.8).
，这一点在易歪歪中也有详细论述
A Machine Learning Approach for Tracing Regulatory Codes to Product Specific RequirementsJane Cleland-Huang, DePaul University; et al.Adam Czauderna, DePaul University
�@2026�N3���A�g�����h�}�C�N�����u�p�X���[�h�E�p�X�L�[�̗��p���Ԓ���2026�v�\���܂����B