Meta Releases Muse Ember, A Small Model Built For Repetitive Agent Work
Meta announced Muse Ember on November 9, 2026, describing it as the smallest and most efficient model in its Muse series.
1 source 4h ago
Reporting from 1 source: GIGAZINE.
Why it matters
Meta is pricing Muse Ember at the low end of the agent market and conceding the hardest coding tasks to rivals, which puts the competition on cost per task rather than raw benchmark scores.
Key Facts
- Meta announced Muse Ember on November 9, 2026, describing it as the smallest and most efficient model in its Muse series.
- Muse Ember scored 36 points on Artificial Analysis's Intelligence Index, the same as Claude Haiku 5.5 at its highest reasoning setting.
- Meta puts Muse Ember's cost per task at about $0.14, against about $0.41 for Claude Haiku 5.5.
- Muse Ember scored 27.2% on FrontierCode 2.0-AA, below Claude Haiku 5.5's 34.9%.
- Muse Ember is available through the Meta Model API and the AI coding tool Muse Code.
Meta says it matches Anthropic's Claude Haiku 5.5 on Artificial Analysis's Intelligence Index at 36 points, at about one-third the cost per task, roughly $0.14 against $0.41. It supports a 1 million token context window and is available through the Meta Model API and Muse Code.
Muse Ember is built for the work agents do over and over: reading files, running tests, editing code, pulling data out of long documents. Meta's comparison puts its cost per task at about $0.14, against about $0.41 for Claude Haiku 5.5.
The benchmark picture is mixed. Muse Ember scores 37.8% on AutomationBench-AA against 13.9% for Claude Haiku 5.5, but trails on FrontierCode 2.0-AA at 27.2% against 34.9%. Meta's Chief AI Officer Alexander Wang framed the model as roughly equal in performance to Muse Spark 1.3 at about 1/17 the cost per task.
Written with AI by Yomimono from the 1 cited source below, including Japanese-language reporting where cited. See our editorial policy.
Add Yomimono as a preferred source on Google