Meta launches Muse Ember for high-volume agent tasks
Meta Superintelligence Labs has released Muse Ember, a hosted multimodal reasoning model designed for repetitive work inside AI agents. Meta targets it at tool calls, coding subtasks, routing, extraction and long-document parsing. It is the smallest model in the Muse line, positioned below Muse Wonder and Muse Spark.
Ember supports text, images, audio and video, allowing agents to process screenshots, charts, scanned pages, voice memos and recordings. Its context window holds 1 million tokens, while its maximum output is 512K tokens. The model is available through the Meta Model API and Muse Code, and its weights remain closed, preventing local deployment.
At maximum reasoning effort, Ember scored 36 on the Artificial Analysis Intelligence Index, matching Claude Haiku 5.5 and exceeding GPT-6 Luna’s 28. Meta reports 37.8 on AutomationBench-AA and 40.9 on Terminal-Bench 4.0, but 27.2 on FrontierCode 2.0-AA, below Haiku’s 34.9. Input costs $0.10 per million tokens and output costs $0.50. Meta estimates $0.136 per index task, compared with $0.411 for Haiku.
Written by Sakhanda Wire AI from reporting by MarkTechPost. Facts, quotes and full context are in the original article below.