Meta Superintelligence Labs has released Muse Ember, its smallest and most efficient model. Meta Muse Ember is a hosted multimodal reasoning model built for the repetitive steps inside agents. Meta lists tool calls, coding subtasks, routing, extraction and long document parsing as targets. It scores 36 on the Artificial Analysis Intelligence Index, level with Claude Haiku 5.5 at max effort. Meta puts its cost at about $0.14 per index task.
TL;DR
- Size: 1M-token context window, up to 512K output tokens.
- Runs on: Meta’s hosted infrastructure only (Meta Model API and Muse Code). Weights are private, so there is no local GPU, laptop or quantized option.
- Performance: Ties Claude Haiku 5.5 (max) at 36 on the AA Intelligence Index, ahead of GPT-6 Luna (28).
- Best: AutomationBench-AA at 37.8, versus 20.2 for GPT-6 Luna and 13.9 for Claude Haiku 5.5.
- Worst: FrontierCode 2.0-AA at 27.2, behind Claude Haiku 5.5 at 34.9.
- Bottom line: Best: Haiku-level intelligence at $0.136 per task versus $0.411. Worst: closed weights and weaker on mergeable real-repo code.
Muse Ember is a small reasoning model from Meta for high-volume agent subtasks. It is the fourth model under the Muse brand, after Spark, Glimmer and Wonder.
Meta positions it below its larger models. Muse Wonder handles the hardest problems and Muse Spark covers coding and agentic workflows. Ember takes repeatable coding and agent tasks at scale. Meta’s own framing: Muse Wonder plans, Muse Ember handles the volume.
How does Muse Ember work?
Muse Ember is the first model Meta trained using Muse Wonder Full, its largest model.
The model natively accepts text, images, audio and video. That lets an agent read screenshots, charts, scanned pages, voice memos and meeting recordings without a separate model. It also operates desktop apps from screenshots.
The 1M-token context window fits a full codebase or document set in one call. Output runs up to 512K tokens, 4x the 128K cap on both rivals.
How does Muse Ember perform on benchmarks?
All results below are at max reasoning effort, as reported by Meta. Some come from Artificial Analysis (AA), others from Meta’s internal suite.
- Agents: On AutomationBench-AA, 657 Zapier business workflows, Ember completes 37.8% of tasks. GPT-6 Luna reaches 20.2% and Claude Haiku 5.5 reaches 13.9%. On Agents’ Last Exam (internal), Ember scores 61.6 versus 57.1 and 57.0.
- Coding: Ember leads Terminal-Bench 4.0 with 40.9 versus Haiku’s 39.2 and Luna’s 16.4. It also leads the AA Coding Agent Index, 46 to 37 to 31. It loses FrontierCode 2.0-AA, where Haiku scores 34.9 and Ember 27.2.
- Multimodal: Ember ties GPT-6 Luna on GDP.pdf-AA at 22.8. It edges Haiku on OSWorld 2.1, 69.5 versus 68.9 on partial credit. PerceptionBench with tools gives Ember 80.6, Haiku 78.6 and Luna 62.8.
‘;});
box.innerHTML=h;setTimeout(function(){box.querySelectorAll(‘.fil’).forEach(function(f){f.style.width=f.dataset.w+’%’})},40);setTimeout(post,80);}
cbs.forEach(function(b){b.addEventListener(‘click’,function(){bench(b.dataset.c)})});
var P=[[“Muse Ember”,”xhigh”,0.116,33.9,”#0866FF”],[“Muse Ember”,”max”,0.136,35.8,”#0866FF”],
[“Claude Haiku 5.5″,”medium”,0.092,24.6,”#7A8AA3″],[“Claude Haiku 5.5″,”high”,0.159,29.6,”#7A8AA3″],[“Claude Haiku 5.5″,”xhigh”,0.248,34.3,”#7A8AA3″],[“Claude Haiku 5.5″,”max”,0.411,36.2,”#7A8AA3″],
[“GPT-6 Luna”,”high”,0.073,21.4,”#B8C3D4″],[“GPT-6 Luna”,”xhigh”,0.104,23.3,”#B8C3D4″],[“GPT-6 Luna”,”max”,0.152,27.8,”#B8C3D4″]];
var sc=R.querySelector(‘#mtpScat’),tip=R.querySelector(‘#mtpTip’),built=false;
function scat(){if(built){return}built=true;var h=””;
[0,0.1,0.2,0.3,0.4].forEach(function(x){h+=’$’+x.toFixed(2)+’‘});
[20,25,30,35,40].forEach(function(y){h+=’‘+y+’‘});
P.forEach(function(p,i){h+=’‘});
sc.innerHTML=h;var pts=sc.querySelectorAll(‘.pt’);
pts.forEach(function(el,i){setTimeout(function(){el.classList.add(‘in’)},120*i);
function sh(){var p=P[i];tip.innerHTML=’‘+p[0]+’ (‘+p[1]+’): index ‘+p[3]+’ at $’+p[2].toFixed(3)+’ per task’}
el.addEventListener(‘mouseenter’,sh);el.addEventListener(‘click’,sh);});}
R.querySelectorAll(‘.sp’).forEach(function(s){s.addEventListener(‘click’,function(){s.classList.toggle(‘op’);setTimeout(post,320)})});
window.addEventListener(‘load’,post);window.addEventListener(‘resize’,post);setTimeout(post,300);
})();
