Inference
EN

Muse Spark 1.2 vs Muse Spark Open

Change models
Muse Spark 1.2Meta
Muse Spark OpenMeta
IndexAll categories, out of 100.
64.9 ±3.3
65.1 ±3.9 best
GapIndex points below the top.
−0.2 · no clear gap
Top
CoverageIndex weight that has results.
95% best
90%
SpeedOutput tokens per second.
123/s best
86/s
EloLMArena rating and rank.
1484
—
ReleasedMonth of release.
2026-08
2026-10
Input / 1MUS dollars per 1M input tokens.
$1.25
$1.25
Output / 1MUS dollars per 1M output tokens.
$4.25
$4.25
1K chats2K tokens in, 500 out, each.
$4.63
$4.63
1K documents20K tokens in, 1K out, each.
$29.25
$29.25
1K agent tasks200K tokens in, 20K out, each.
$335.00
$335.00
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
62.7
64.8 best
CodingCode writing and repair.
64.3 best
63.3
ReasoningLogic problems and puzzles.
64.1 best
62.3
MultimodalTasks with images and text.
67.2
69.8 best
KnowledgeFacts and expert knowledge.
65.2
65.4 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
77.0
—
MathMath problems.
64.4
64.9 best

Benchmarks

24 shared
Only Muse Spark 1.2: 33Only Muse Spark Open: 21
Muse Spark 1.2
Muse Spark Open
Artificial Analysis Humanity's Last ExamKnowledge
45.5% best27 / 222
43.9%30 / 222
Artificial Analysis Long Context ReasoningReasoning
79.0% best58 / 217
78.3%63 / 217
Critical Physics TasksReasoning
17.7% best37 / 214
14.9%51 / 214
Artificial Analysis GPQA DiamondKnowledge
90.4% best33 / 199
89.0%48 / 199
Artificial Analysis Omniscience AccuracyKnowledge
45.4% best36 / 164
42.0%44 / 164
GDPval-AA normalizedAgentic
48.6% best28 / 133
45.1%39 / 133
Vibe Code Bench v1.1Coding
79.1% best32 / 123
76.2%41 / 123
Artificial Analysis SciCodeCoding
57.4%19 / 114
58.4% best17 / 114
SimpleQA VerifiedKnowledge
60.3% best18 / 91
56.6%22 / 91
Legal Research BenchLegal
43.8% best25 / 90
38.5%39 / 90
Tax Agent BenchFinance
32.8% best20 / 82
26.3%42 / 82
LiveBench Data AnalysisReasoning
76.5% best33 / 80
60.2%72 / 80
LiveBench ReasoningReasoning
90.0% best14 / 80
73.7%71 / 80
Finance Agent (v2)Finance
60.6%4 / 79
61.3% best3 / 79
Code MigrationCoding
30.0% best42 / 77
21.2%51 / 77
EMBFinance
57.0% best42 / 75
53.7%51 / 75
Terminal-Bench 4.0Agentic
6.1%50 / 61
6.6% best49 / 61
DeepSWEAgentic
59.3%43 / 59
60.0% best41 / 59
LiveBench Coding 2.0Coding
78.2% best33 / 52
73.6%37 / 52
LiveBench DesignAgentic
67.1% best41 / 52
60.3%47 / 52