benchgap
Meta · model

Muse Glimmer (high) benchmark scores

As of 2026-10-07, Muse Glimmer (high) (Meta) has measured scores on 11 benchmarks and estimated scores on 3 more.

BenchmarkScoreSource
Terminal-Bench 2.151.7%measured
Terminal-Bench 4.00.5%measured
Terminal-Bench Science 0.12.6%estimated ± 6.5 pp, low confidence
GPQA Diamond83.5%measured
SciCode44.9%measured
CritPt2.6%measured
GDP.pdf10.0%measured
τ³-Bench Banking23.5%measured
MMMU-Pro74.0%measured
AA-AnalystAgent16.2%estimated ± 3.6 pp, medium confidence
Harvey LAB90.1%estimated ± 1.2 pp, medium confidence
Humanity's Last Exam22.0%measured
AutomationBench6.8%measured
EnterpriseOps-Gym34.7%measured