What is Muse Ember? Meet Meta’s new low-cost AI model

Meta has unveiled Muse Ember, its newest AI model designed to be low-cost and highly efficient for repetitive tasks. Unveiled on November 9 by Meta's Superintelligence Labs, Ember is part of a family of AI models under the Muse name. Each model serves a unique purpose - Wonder for complex problems, Spark for coding and tool usage, and Ember for repeatable tasks.

Ember stands out for its affordability and capability to handle large volumes of data. It can process up to a million tokens at once, equipping developers to build agents that can read contracts, analyze meeting recordings, and more. For comparison, Anthropic's Claude Haiku 5.5 scored similarly at a higher cost, while OpenAI's GPT-6 Luna scored slightly lower at a similar price point.

Meta's pitch is primarily about cost-effectiveness. Developers can use Wonder to plan complex tasks and then delegate the individual steps to Ember, ensuring efficient handling of mundane and repetitive work. The model is currently available via Meta's Muse Code tool and developer API.

In a demonstration, Meta's cloud storage partner Box used Ember to review contracts and assign risk ratings based on company guidelines. Ember can handle multiple types of data including images, audio, and video, making it versatile for various applications.

Ember's cost-effectiveness has been highlighted in Meta's benchmark results, where it outperformed Anthropic's Claude Haiku 5.5 at the same price point. Moreover, it has shown potential in terminal coding tests and code merging evaluations. Meta is also reportedly developing even more cost-efficient models.

As AI assistants become more involved in everyday tasks, the need for affordable ways to handle these tasks will grow. Meta's Muse Ember is positioned to fill this gap, making once-expensive tasks more accessible for developers.