Meta Releases Muse Ember, a Lightweight AI Model for Running AI Agents

At a glance

Meta Superintelligence Labs has released Muse Ember, its lightest and most cost-efficient AI model. It is designed for tasks that AI agents (AI programs that handle multi-step work on their own) repeat throughout the day, such as tool calls and code subtasks.

Key points

1. 1

   The company emphasized the cost per task, saying that at 10 cents per one million input tokens (small units of text), it is cheaper than Anthropic's Haiku and OpenAI's Luna.

2. 2

   Ember is designed to be used alongside the larger Muse Wonder model: Muse Wonder sets the rules, and Ember handles most of the actual processing work.

3. 3

   It can handle text, images, and voice together, so it can process screenshots, charts, meeting recordings, and videos without a separate model.

Why it matters

The cost of using AI at scale determines service prices and how quickly it can be adopted.

Also reported by

• The DecoderMeta launches Muse Ember, a small AI model built for the repetitive work agents actually do
• Unite.AIMeta, 에이전트 및 코딩 작업 비용을 낮추는 Muse Ember 출시