benchgap
Meta · model

Muse Ember benchmark scores

As of 2026-11-18, Muse Ember (Meta) has measured scores on 10 benchmarks and estimated scores on 55 more.

BenchmarkScoreSource
AA-GPQA Diamond73.6%estimated ± 11.5 pp, low confidence
AA-HLE39.4%estimated ± 9.2 pp, medium confidence
Artificial Analysis Intelligence Index36.0%measured
CritPt15.5%estimated ± 5.3 pp, medium confidence
GDPval-AA57.6%estimated ± 4.2 pp, high confidence
GPQA Diamond91.9%estimated ± 2.5 pp, high confidence
HLE26.5%estimated ± 10.3 pp, low confidence
AA-Omniscience Accuracy21.1%measured
Terminal-Bench 2.189.1%estimated ± 5.4 pp, low confidence
BrowseComp77.6%estimated ± 11.3 pp, low confidence
HLE w/ tools58.8%estimated ± 4.9 pp, medium confidence
AA-MMMU-Pro82.1%estimated ± 5.5 pp, medium confidence
BioMysteryBench (human-difficult)31.7%estimated ± 7.8 pp, low confidence
AutomationBench47.1%estimated ± 6.1 pp, low confidence
JobBench66.7%estimated ± 11.3 pp, low confidence
MMLU-Pro84.1%estimated ± 5.1 pp, medium confidence
GPQA92.1%estimated ± 6.9 pp, medium confidence
MMLU89.3%estimated ± 0.9 pp, high confidence
Vals GPQA Diamond90.7%estimated ± 5.7 pp, medium confidence
Vals MMLU-Pro79.7%estimated ± 2.9 pp, high confidence
MMMLU92.4%estimated ± 3.0 pp, low confidence
AA MMLU-Pro84.2%estimated ± 0.8 pp, medium confidence
ApprenticeBench13.1%estimated ± 12.5 pp, low confidence
ARC-AGI-163.5%estimated ± 7.9 pp, low confidence
ARC-AGI-264.9%estimated ± 14.6 pp, low confidence
Terminal-Bench 3.027.2%estimated ± 3.2 pp, high confidence
MazeBench0.0%measured
ParseBench-169.1%estimated ± 3.7 pp, high confidence
Terminal-Bench 4.130.6%measured
Terminal-Bench Hard60.9%estimated ± 2.8 pp, medium confidence
Terminal-Bench 2.1 (Vals AI)75.8%estimated ± 6.9 pp, low confidence
CWE-Bench-AA62.6%estimated ± 9.0 pp, medium confidence
DeepsecBench-AA34.0%estimated ± 7.5 pp, medium confidence
Harvey LAB-AA v1.14.3%estimated ± 3.5 pp, low confidence
HealthBench Professional (length-adjusted, Artificial Analysis)56.8%estimated ± 4.3 pp, low confidence
HealthBench Professional (raw, Artificial Analysis)64.9%estimated ± 1.1 pp, low confidence
CWE-bench v157.3%estimated ± 5.3 pp, medium confidence
DRACO1.6%estimated ± 0.7 pp, low confidence
HLE w/o tools44.7%estimated ± 5.0 pp, high confidence
OSWorld 2.169.5%measured
OSWorld 2.1 Offline35.0%estimated ± 13.1 pp, low confidence
Terminal-Bench 4.040.9%measured
Terminal-Bench-Science 0.115.1%estimated ± 10.2 pp, low confidence
Toolathlon-Verified73.4%estimated ± 2.6 pp, high confidence
AA AutomationBench68.1%measured
GDP.pdf22.8%measured
AA Terminal-Bench 4.040.9%measured
CWE-bench54.1%estimated ± 10.3 pp, low confidence
MKA46.2%estimated ± 5.3 pp, medium confidence
Hermes Bench71.8%estimated ± 1.8 pp, medium confidence
Hermes Index37.4%estimated ± 2.7 pp, medium confidence
ParseBench-275.5%measured
ExploitGym12.7%estimated ± 6.4 pp, medium confidence
C-Eval96.3%estimated ± 1.0 pp, low confidence
MMLU-Redux93.6%estimated ± 0.8 pp, medium confidence
MMMU-Pro81.8%estimated ± 5.4 pp, low confidence
SuperGPQA64.9%estimated ± 9.3 pp, medium confidence
Finance Agent v256.1%estimated ± 3.8 pp, medium confidence
HealthBench (length-adjusted)54.1%estimated ± 3.7 pp, high confidence
HealthBench Professional57.4%estimated ± 4.2 pp, medium confidence
HLE-Verified16.4%estimated ± 7.1 pp, low confidence
STEMBench2.3%estimated ± 0.2 pp, low confidence
SimpleQA19.6%estimated ± 4.3 pp, medium confidence
Agents' Last Exam35.3%estimated ± 10.8 pp, low confidence
BioBasicsBench34.1%estimated ± 3.8 pp, low confidence