Muse Ember

Muse Ember is Meta's smallest and most efficient reasoning model, built for the work agents repeat all day: tool calls, coding subtasks, routing, extraction, and long-document parsing. It pairs with Muse Wonder or Muse Spark, which plan while Muse Ember handles the volume.

Strongest at
Reading & documents
Top 11% of the past year's models · 2 of 4 benchmarks
Context window
1.05M
tokens in · 524.3K out

Capability profile

Scored against 92 models released in the past year. 50 is the average model; further out is better.

This modelMedian of the past year
50100Median of the past yearReading & documents: 89 · Top 11%Reasoning: 70 · Top 28%Visual reasoning: 90 · Top 16%Computer use: 95 · Top 19%ReadingReasoningVisionComputer use
Reading & documents89Top 11%
Reasoning70Top 28% · 1 test
Visual reasoning90Top 16%
Computer use95Top 19% · 1 test

No results yet: Coding, Writing, Math, Science

Where these scores come from

Each capability combines the published benchmarks listed here, rescaled so 50 is the reference average and 15 points is one standard deviation. A hollow dot rests on a single benchmark.

Epoch benchmark boards

Best result on each board it has run. The tick marks the leader.

SimpleBench
73%#26 of 117
APEX-Agents
64%#21 of 65

Meta's past year

9 release days in the past year, about one every 41 days. The next is expected around Dec 31, 2026. Forecasts ↗

DecJanFebMarAprMayJunJulAugSepOctNovDec

Specifications

RELEASE DATENov 9, 2026
WEIGHTSClosed weights
CONTEXT WINDOW1.05M
MAX OUTPUT524.3K
REASONINGNot listed
TOOL CALLINGNot listed
INPUTStext, image, audio, video, pdf
OUTPUTStext

Models.dev record ↗

Specifications and prices from Models.dev; availability and limits may differ by provider. Benchmarks from Epoch AI (CC BY 4.0), LiveBench and the other sources named above.