Skip to content
Hot storiesContinuously updated

Meta发布智能体执行模型Muse Ember

4 reports4 reporting sourcesUpdated 3 hours ago

Start here

AI overview

2026年11月10日,Meta Superintelligence Lab 发布轻量级多模态模型 Muse Ember,面向智能体执行任务。该模型原生支持文本、图像与音频输入,输入价格为每百万 token 0.10 美元,适用于工具调用、简单编码、请求路由、数据提取与长文档解析等重复性任务;它常与前沿大模型 Muse Wonder 搭配,由 Wonder 制定规则,Ember 承接执行。模型可通过 Muse Code 和 Meta Model API 使用,标识为 muse-ember-1.0。最新报道补充称,该模型支持 100 万上下文,专为端到端智能体工作流与复杂推理设计,官方称其表现媲美 Claude Haiku 5.5,但单任务成本仅为后者的三分之一。

AI-generated from reports · Updated 3 hours ago

Report timeline

Follow the reports to see different sides of the story.

Nov 10
  1. IT之家 · AI 筛选Selected
    Meta launches lightweight model Muse Ember: Features 1M context window at one-third the per-task cost of Haiku 5.5

    Meta has introduced Muse Ember, a lightweight multimodal model engineered for end-to-end agentic workflows and complex reasoning. Meta claims it delivers performance comparable to Claude Haiku 5.5 at just one-third of the per-task cost.

  2. Buzzing HN · 中文精选Selected
    Meta releases Muse Ember for efficient agentic and coding tasks

    Meta has released Muse Ember for repetitive agentic tasks such as tool calling, coding subtasks, routing, and long-document parsing. It is available through Muse Code and the Meta Model API under the model identifier muse-ember-1.0.

  3. The DecoderSelected
    Meta releases small AI model Muse Ember for repetitive agent tasks

    Meta has released Muse Ember, its smallest and most efficient AI model to date, designed for agent tool calling, simple coding, request routing, data extraction, and long-document processing.

  4. TechCrunch AISelected
    Meta Releases Lightweight Agent Execution Model Muse Ember

    Meta Superintelligence Lab has released Muse Ember, a lightweight and cost-effective AI model designed for agent execution tasks such as tool calling, code subtasks, and long-document parsing. Priced at $0.10 per million input tokens, it natively supports multimodal inputs across text, images, and audio. It is typically paired with the frontier model Muse Wonder, where the flagship model sets instructions while Ember handles the processing and execution of high-volume routine tasks.

Story heat trend

Current heat 33·Comparable peak 34(Nov 10 10:00)·Comparable change, last 24 hours–

010203040Nov10Nov10Nov10Nov10

Trends compare only the same parties observed continuously, so the scope can be narrower than current heat counts. Move the pointer or tap the chart to see hourly heat; use the left and right arrow keys on a keyboard.