Meta AImodel systemcard
Muse Spark
#5of 27 · general
1.0 and 1.1 collapsed
Standing
dot = this model · shaded = the field's middle 60% · track = full fieldGeneral
73.2#5/27
Math
43.9#13/27
Coding
69.7#5/27
Health
87.5#1/26
Engineering
75.0#5/17
Legal
82.4#1/26
Instruct
59.8#9/27
Agents
73.0#4/27
Professional
83.0#4/27
Multimodal
58.4#6/19
Hallucination
51.1#5/27
Value
69.5#2/27
Measured
≈ imputed · amber dot = lab-self-reported among the sourcesGPQA Diamond90.1%#19
HLE45.5%#9
MMLU-Pro88.7%#9
AIME 202596.9%#3
LiveBench75.3%#10
SWE-bench V79.7%#8
LiveCodeBench85.9%#13
SciCode57.3%#4
CritPt16.4%#12
Terminal-Bench69.3%#14
APEX Agents41.9%#3
GDPval56.4%#6
Arena1491#3
AA Index56.8#6
Vals Index68.4#7
Epoch Capabilities Index154.8#11
Operations
$1.25 / $4.25$ per 1M tokens, in / out
$2blended
$0.19$ per solved task
1491arena elo · 9,941 votes