Meta launches Muse Ember, its smallest model, matching Claude Haiku 5.5 at ~1/3 the cost

AIatMeta · x · 2026-11-10

Meta introduced Muse Ember, its smallest and most efficient model, built for work agents repeat all day: tool calls, coding subtasks, routing, extraction, and long document parsing.

Per Artificial Analysis' Intelligence Index, it beats Claude Haiku 5.5 and GPT-6 Luna on intelligence per dollar at every reasoning setting: at max effort it scores 35.8 at $0.116–0.136 per task, versus Haiku 5.5's 36.2 at $0.411 and Luna's 27.8 at $0.152. It sits on the efficiency frontier of the full leaderboard. Available now via Meta's API.

Related event: Meta Launches Muse Ember, Its Smallest Model Built for Agentic Value(17 posts)→

Original post →

More from Models

Models channel →