Introducing Our Smallest, Most Efficient Model: Muse Ember
Summary
Meta introduces Muse Ember, the smallest and most efficient model designed for routine agent work, coding subtasks, and document parsing. It claims better intelligence-per-dollar than Claude Haiku 5.5 and GPT-6 Luna across various reasoning tasks and offers a favorable cost model and large context window. Availability is through Muse Code and Meta Model API with pricing suitable for scalable enterprise use.