←新闻详情

Rohan Paul更新于 10月21日 22:46

Meta 发布 Muse Wonder,多项基准超过 Opus 5.5 与 GPT-6 Astra

Meta is so back. in each of these benchmarks Muse Wonder beats both Opus 5.5 and GPT-6 Astra

AI 导读

Meta 发布新前沿模型 Muse Wonder,官方称其在个人智能体和专业智能体任务上达到 SOTA。Rohan Paul 指出,Muse Wonder 在多项基准上均超过 Opus 5.5 和 GPT-6 Astra,差距最大的是 SWE-Marathon,该基准要求 AI 智能体在 2 到 10 小时内独立完成整个软件项目(如 C 编译器或 Slack 克隆),且所有隐藏检查通过才算完成;Muse Wonder full 在该项得分为 57.5,Opus 5.5 为 46.3。

推荐理由

AI 评分 67

Meta 新模型 Muse Wonder 在多项基准上超过 Opus 5.5 与 GPT-6 Astra,其中 SWE-Marathon 差距最大,可据此观察长时程智能体任务的能力对比。

主题

标签

引用

Alexandr Wang

@alexandr_wang

introducing muse wonder our new frontier model. state-of-the-art at personal agents and professional agents. our largest and most capable model yet—rivals the big bois https://t.co/REYHXbjpf6

查看原帖

正文

当前显示原文,中文全文译文尚未就绪。

Meta is so back. in each of these benchmarks Muse Wonder beats both Opus 5.5 and GPT-6 Astra

The widest gap comes on SWE-Marathon. this one tests whether an AI agent can build an entire software project on its own, such as a C compiler or a Slack clone, over 2 to 10 hours, and counts a task only if every hidden check passes.

And muse wonder full scores 57.5 against 46.3 for Opus 5.5.

来源

Rohan Paul