Meta Superintelligence Lab on Monday introduced a new AI model named Muse Ember, which the lab describes as its lightest and most cost-efficient model to date. According to Meta, Ember was built for complex reasoning execution and the day-to-day work agents repeat: tool calls, coding subtasks, routing, extraction, and long-document parsing.
Cost and benchmark claims
Meta highlighted Ember’s cost-per-task, citing Artificial Analysis’s composite benchmark score to position Ember as both cheaper and more capable than competitors such as Anthropic’s Haiku and OpenAI’s Luna. The company states that Ember’s input-token price is $0.10 (10 cents) per one million input tokens, which it says is notably lower than frontier competitors.
Meta also claims Ember is cost-competitive with open-source alternatives like DeepSeek V4.1 Flash and GLM 5.3 Flash.
Designed to work with larger models at scale
Given the emphasis on low cost and accuracy at scale, Meta envisions Ember paired with a larger model such as Muse Wonder in agentic deployments. For example, in a customer-support scenario a heavier model could generate rules for handling a particular issue, and a lighter-weight model like Ember would apply those rules across tens of thousands of cases. As Meta put it in its announcement: “Muse Wonder plans, Muse Ember handles the volume.”
Multimodal input handling
With high-volume use cases in mind, Meta trained Ember to handle multiple media types, including text, images and audio. That training makes the model flexible in the kinds of inputs it can process; Meta says agents can read screenshots, charts and scanned documents, or work from voice memos, meeting recordings and video without requiring a separate model.
Position within the Muse family
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 frontier Wonder model (released in October). Meta’s flagship consumer AI product, the Muse Agent, was released in September and — according to the company — achieved unexpected success; features such as Spaces and video chat rolled out earlier this month.
Why this matters
Muse Ember represents a strategic move by Meta to combine larger planning models with smaller, low-cost execution models for agent-style workloads. Its lower price point and multimodal capabilities make it a candidate for enterprise, customer-support, and other volume-heavy automation tasks where scalability and cost-efficiency are key considerations.
