Meta Superintelligence Lab on Monday released Muse Ember, its lightest and most cost-efficient AI model to date. The model is designed for complex reasoning execution and the repetitive work agents do all day: tool calls, coding subtasks, routing, extraction, and parsing long documents. Meta is positioning Ember around cost per task as much as raw model capability.
Meta used Artificial Analysis’s composite benchmark score to present Ember as cheaper and smarter than Anthropic’s Haiku and OpenAI’s Luna. At 10 cents per million input tokens, Ember is significantly cheaper than those frontier competitors, and Meta said it is cost-competitive with open-source rivals DeepSeek V4.1 Flash and GLM 5.3 Flash. The comparison centers on the economics of running agents at scale.
Ember is built to pair with a larger model, such as Meta’s Muse Wonder, inside an agentic harness. In a customer-support agent, for example, a heavier model could establish rules for handling a particular issue and then send Ember to apply those rules across tens of thousands of cases. Meta summed up the split: "Muse Wonder plans, Muse Ember handles the volume."
With volume-heavy use cases in mind, Meta trained Ember to work across text, images, and audio. Meta described the model as surprisingly flexible in the inputs it can handle. Meta said agents can read screenshots, charts, and scanned documents, or work from voice memos, meeting recordings, and video without a separate model.
Ember is the fourth model released under Meta’s Muse brand. It follows the mid-weight Spark, released in April; Spark’s open-source twin Glimmer, released in August; and the Wonder frontier model, released in October. Meta’s flagship consumer AI product, the Muse Agent, launched in September to unexpected success, with features such as Spaces and video chat rolling out earlier this month.
The release extends the Muse lineup toward cheaper execution, with larger models setting strategy while Ember absorbs the high-volume, repeated steps that make agent products expensive to run.