Meta launches Muse Ember, its cheapest model, built for high-volume agent work

Meta Superintelligence Labs released Muse Ember on 9 November 2026, calling it its smallest and most efficient AI model so far. Meta says it was designed for reasoning execution and “the work agents repeat all day”: tool calls, coding subtasks, routing, extraction and long document parsing.

Price and positioning

• Input price: 10 cents per million tokens, according to TechCrunch — well below Anthropic’s Haiku and OpenAI’s Luna, and competitive with open models such as DeepSeek V4.1 Flash and GLM 5.3 Flash.
• Meta claims Ember gives “more intelligence per dollar” than Claude Haiku 5.5 or GPT-6 Luna at any reasoning setting, citing Artificial Analysis’s composite benchmark. This is a vendor comparison, not an independent test.
• Ember is meant to work alongside the larger Muse Wonder in agent systems. Meta’s summary: Wonder plans, Ember “handles the volume” — for example, applying rules a bigger model wrote to tens of thousands of customer-support cases.

Multimodal

Meta says Ember accepts 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.

The Muse family

Ember is the fourth model under the Muse brand, after mid-size Spark (April), its open-source twin Glimmer (August) and frontier Wonder (October). The consumer Muse Agent launched in September.