Meta Superintelligence Lab released a new artificial intelligence model on Monday named Muse Ember, positioning it as the lab's lightest and most cost-efficient AI model yet. In its official release, Meta stated that Ember was engineered specifically for complex reasoning execution and the repetitive workflows that AI agents handle throughout the day, including tool calls, coding subtasks, routing, data extraction, and long document parsing.
Benchmark Performance And Cost Efficiency
Meta placed a strong emphasis on Ember's cost-per-task metric, utilizing Artificial Analysis's composite benchmark score to demonstrate that Ember outperforms competitors like Anthropic’s Haiku and OpenAI’s Luna while remaining significantly cheaper. Priced at 10 cents per million input tokens, Ember undercuts frontier rivals substantially and remains highly competitive against open-source alternatives such as DeepSeek V4.1 Flash and GLM 5.3 Flash.
Pairing With Larger Frontier Models
Given its focus on affordability and high-scale accuracy, Ember is structured to function alongside a larger foundational model, such as Meta’s Muse Wonder, within an agentic harness. For instance, in a customer support environment, a heavier model might establish the underlying rules for handling specific grievances, while a lightweight model like Ember executes those rules across tens of thousands of individual cases. As Meta described it in the announcement, Muse Wonder plans while Muse Ember handles the volume.
Multi-Modal Capabilities And Product Lineup
To support high-volume use cases effectively, Meta trained Ember to process diverse media formats, encompassing text, images, and audio. This flexibility allows agents to interpret screenshots, charts, and scanned documents, or operate directly from voice memos, meeting recordings, and video feeds without relying on a separate auxiliary model. Ember represents the fourth model unveiled under the Muse brand, following the mid-weight Spark released in April, Spark's open-source counterpart Glimmer released in August, and the Wonder frontier model introduced in October. Meanwhile, Meta's flagship consumer offering, the Muse Agent, launched in September to unexpected success, introducing collaborative features like Spaces and video chat earlier this month.
