Meta 发布 Muse Wonder 系列模型，主打个人与专业智能体任务
The Muse Wonder series achieves state-of-the-art results across key benchmarks for personal and professional agents, memory, and coding.
Meta 发布 Muse Wonder 和 Muse Wonder Full 两款前沿模型，称其在个人与专业智能体、记忆和编码等关键基准上取得 SOTA。Muse Wonder 即日起通过 Meta Model API 和 Muse Code 提供，Muse Wonder Full 面向 Muse Code 订阅用户；Muse Wonder 还将接入个人智能体产品 Muse，先向订阅用户开放。两款模型引入 Active Memory 跨对话保留上下文，以及 Active Focus 在任务受阻时尝试替代方案并坚持完成目标。
推荐理由
精选AI 评分 67
Meta 官方给出 Muse Wonder 系列的基准表现、可用入口与记忆机制，读者可据此判断其个人智能体定位。
主题
正文
当前显示原文，中文全文译文尚未就绪。
The Muse Wonder series achieves state-of-the-art results across key benchmarks for personal and professional agents, memory, and coding.
Muse Wonder and Muse Wonder Full apply specialized knowledge to problems people encounter at work and in daily life, from understanding a lease to making sense of financial or health information. Our evaluations include tasks in law, finance, and health that test this ability.