Meta Releases Muse Ember, Its Lightest and Cheapest AI Model

Meta Superintelligence Lab released Muse Ember on Monday, its lightest and most cost-efficient AI model yet, priced at 10 cents per million input tokens and built for the high-volume work that AI agents repeat all day.

In its release, Meta said Ember was designed for complex reasoning execution and the work agents repeat all day, naming tool calls, coding subtasks, routing, extraction and long document parsing. The lab used Artificial Analysis's composite benchmark score to show Ember as cheaper and smarter than competitors such as Anthropic's Haiku and OpenAI's Luna, and cost-competitive with open-source models such as DeepSeek V4.1 Flash and GLM 5.3 Flash.

Ember is designed to pair with a larger model in an agentic harness. As Meta put it in the announcement, Muse Wonder plans and Muse Ember handles the volume. In a customer-support agent, for example, a heavier model could set the rules for handling an issue while Ember applies those rules across tens of thousands of cases.

Ember was also trained to work across text, images and audio, so agents can read screenshots, charts and scanned documents or work from voice memos, meeting recordings and video without a separate model, Meta said.

Ember is the fourth model released under Meta's Muse brand, following the mid-weight Spark in April, Spark's open-source twin Glimmer in August and the Wonder frontier model in October. The flagship consumer AI product, the Muse Agent, was released in September to unexpected success, with new features such as Spaces and video chat rolling out earlier this month.

Meta (META) shares were up 5.1% in delayed trading.