Meta Releases Muse Ember, a Small Efficient Model for Agentic Workloads
Meta AI launched Muse Ember, described as its smallest and most efficient model, targeting complex reasoning execution and repetitive agent work such as tool calls, coding subtasks, routing, extraction, and long document parsing. For shorter prompts it is priced at $0.10 per million input tokens and $0.50 per million output tokens, matching GPT-6 Luna and Claude Haiku 5.5, with higher rates above a 200k-token threshold. The model offers six reasoning levels (minimal through max) and, according to Simon Willison's write-up, sits on the Pareto frontier of Artificial Analysis's intelligence-versus-cost-per-task chart. Willison also noted that reasoning level did not map cleanly to output quality in his SVG pelican test.
Why it matters: It gives readers concrete details on a newly released Meta model, including pricing tiers, positioning against competing small models, and efficiency benchmarks.