Muse Ember

Nov 9, 2026Former stealth modelReasoningClosed weightsModels.dev

Intelligence: 36.0; Rank 75 of 782 measured models
Output speed: 448 t/s; Rank 4 of 449 measured models
Context window: 1.05M; Rank 49 of 1325 measured models
Cost per task: $0.136; Rank 57 of 229 measured models

Muse Ember (Max) scores 36.0 on the Artificial Analysis Intelligence Index, ranking 75th of 782 measured models. The closest models on this index are Claude Haiku 5.5 (36.5), Agnes 3.0 Flash (35.5), and Grok 4.7 (35.5). With an output speed of 448 t/s, it is faster than Claude Haiku 5.5 (255 t/s) and Grok 4.7 (63.3 t/s).

Stealth model history

The provider was anonymous during this preview. The identities below are confirmed by OpenRouter. First seen is our first observation, not the model release date.

Crouton Alpha stealth/crouton-alpha
Listed on OpenRouter
Nov 3, 2026
First seen by BenchSift
Nov 3, 2026
Identity revealed
meta/muse-ember
OpenRouter source

Artificial Analysis Indices

Maximum

Intelligence36.0

Reasoning levels

Each point is a reasoning level of this model: more reasoning often raises the score but costs more. Select a point to open that level.

Capability indexes

Performance in specific industries, measured by Artificial Analysis

Finance & Accounting44.6

Strategy & Ops48.7

Legal38.5

Healthcare & Medical41.3

Engineering41.8

Economics41.0

Standard Benchmarks

FrontierCode 2.027.2%

Terminal-Bench-Science20.0%

HLE-Diamond25.1%

CritPt v2 (Hard Subset)39.0%

AutomationBench-AA68.1%

GDP.pdf (all-pass)22.8%

Terminal-Bench 4.040.9%

AA-Omniscience21.1%

Models at a similar level

The closest models on the Intelligence Index, one configuration per family.

Muse Spark 1.3
Meta · $2 · 138 t/s
36.6+0.6
$2138 t/sCompare
Claude Haiku 5.5
Anthropic · $0.20 · 255 t/s
36.5+0.5
$0.20255 t/sCompare
Agnes 3.0 Flash
Sapiens AI · $0.075 · —
35.5−0.5
$0.075—Compare
Grok 4.7
SpaceXSI · $3 · 63.3 t/s
35.5−0.5
$363.3 t/sCompare