Inference
EN
Meta

Muse Spark 1.2

Muse Spark 1.2 is a reasoning model from Meta in the Muse Spark family. 34 benchmarks count toward its index, in 7 categories.

  • Closed weights
  • Reasoning
  • Input: text · image · video · file
  • Released 2026-08
RankedEnough results for an index.
IndexAll categories, out of 100.64.9±3.3Median 51.0
CoverageIndex weight that has results.95%
SpeedOutput tokens per second.123/sMedian 67/s
First tokenSeconds to the first output token.3.7 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$1.25Median $0.60
Output / 1MUS dollars per 1M output tokens.$4.25Median $2.55
ContextMaximum tokens in one request.1.05MMedian 262K
EloLMArena rating and rank.1484±6 #17Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
meta/muse-spark-1.2
Max outputMaximum output tokens per request.
944K
Knowledge cutoffLast date of training data.
—
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.15
WeightsLink to the published weights.
—
RetiresDate OpenRouter removes the model.
—

Where to run

APIs that serve this model.
2 providers
ProviderTypeMaker, cloud, host or router.InputOutputContext
MetaMaker$1.25$4.251.05M
OpenRouterRouter$1.25$4.251.05M

US dollars per 1M tokens, list price. From models.dev (opens in a new tab) (MIT).

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.9 counted62.757 / 218
CodingCode writing and repair.11 counted64.347 / 209
ReasoningLogic problems and puzzles.4 counted64.150 / 227
MultimodalTasks with images and text.1 counted67.2—
KnowledgeFacts and expert knowledge.6 counted65.244 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted77.0—
MathMath problems.2 counted64.433 / 168

Results

34 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
Artificial Analysis Humanity's Last ExamKnowledge45.5%27 / 222
IOI v1Coding49.5% ±3.79 / 61
MMLU ProKnowledge88.3% ±0.320 / 135
SWE-benchCoding86.6% ±1.513 / 87
Artificial Analysis GPQA DiamondKnowledge90.4%33 / 199
LiveBench Instruction FollowingInstruction74.3%11 / 66
Artificial Analysis SciCodeCoding57.4%19 / 114
Critical Physics TasksReasoning17.7%37 / 214
LiveBench ReasoningReasoning90.0%14 / 80
SimpleQA VerifiedKnowledge60.3% ±1.618 / 91
GDPval-AA normalizedAgentic48.6%28 / 133
Artificial Analysis Omniscience AccuracyKnowledge45.4%36 / 164
MMMU ProMultimodal86.1% ±0.822 / 92
Vibe Code Bench v1.1Coding79.1% ±3.332 / 123
Artificial Analysis Long Context ReasoningReasoning79.0%58 / 217
Agent Arena command recoveryAgentic5.814 / 50
VulcanBench v3Coding87.0%4 / 14
Terminal-Bench 2.1 · BenchLMAgentic82.9%26 / 86
LiveBench Agentic CodingAgentic57.6%20 / 66
Terminal-Bench 2.1 · Vals AIAgentic69.7% ±0.628 / 75
LiveBench MathematicsMath91.2%27 / 66
LiveBench Data AnalysisReasoning76.5%33 / 80
MazeBench: visual spatial reasoning in a 3D open worldAgentic0.0%23 / 48
SkillsBenchCoding53.0% ±4.417 / 35
LiveBench CodingCoding77.5%35 / 66
LiveBench LanguageKnowledge78.6%36 / 66
Code MigrationCoding30.0% ±4.042 / 77
Chess PuzzlesReasoning27.0% ±4.552 / 90
LiveBench Coding 2.0Coding78.2%33 / 52
ProofBench v1.1Math43.0% ±5.036 / 54
LiveBench Browser UseAgentic36.0%35 / 49
DeepSWEAgentic59.3%43 / 59
LiveBench ResearchAgentic67.7%39 / 52
LiveBench Long AgenticAgentic21.3%39 / 51
LiveBench DesignAgentic67.1%41 / 52
Agent Arena task outcomeAgentic−9.843 / 54
Terminal-Bench 4.0Agentic6.1% ±1.550 / 61
Agent Arena steerabilityAgentic−12.547 / 54
IOICoding21.8% ±0.942 / 45
Artificial Analysis Terminal-Bench v2.1Agentic80.1%20 / 21
FrontierSWE v2Coding12.0%21 / 22
TaxEval v2Finance80.4% ±0.81 / 143
CorpFin v2Finance70.9% ±0.95 / 131
MedScribeHealthcare90.1% ±2.05 / 109
Finance Agent (v2)Finance60.6% ±0.34 / 79
LegalBenchLegal85.3% ±0.422 / 149
Public Benefits Bench v1.1Public benefits68.5% ±1.210 / 50
MedCodeHealthcare49.3% ±2.224 / 106
Tax Agent BenchFinance32.8% ±2.820 / 82
Legal Research BenchLegal43.8% ±3.425 / 90
ProgramBenchCoding0.5% ±0.519 / 65
MortgageTaxFinance65.4% ±0.943 / 97
CyberBench v1.1Other69.5% ±5.624 / 48
EMBFinance57.0% ±3.142 / 75
BioMysteryBenchScience64.8% ±1.323 / 30
SRE BenchOther0.0%33 / 36
Artificial Analysis EnterpriseOps-GymAgentic47.3%—

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseLiveBenchNo licence statedVals AINo licence statedEpoch AICC BYLMArenaCC BY 4.0models.devMIT

History

5 releases
Index66.2 → 64.9
23 Oct 202612 Nov 2026
Output / 1M$4.25 → $4.25
23 Oct 202612 Nov 2026

Questions

How much does Muse Spark 1.2 cost?
$1.25 per 1M input tokens, $4.25 per 1M output tokens.
How good is Muse Spark 1.2?
Its index is 64.9 out of 100. That is place 45 of 283 ranked models.
What is Muse Spark 1.2 best at?
Knowledge: 65.2 out of 100, place 44 of 268.
How fast is Muse Spark 1.2?
123 output tokens per second, 3.7 s to the first token.
What is the context window of Muse Spark 1.2?
1.05M tokens.
Is Muse Spark 1.2 open source?
No. Its weights are not published.
When was Muse Spark 1.2 released?
August 2026.

Badge

Muse Spark 1.2 on Inference360: place 45 of 283, index 64.9
HTML
<a href="https://inference360.ai/models/muse-spark-1-2"><img src="https://inference360.ai/badges/muse-spark-1-2.svg" alt="Muse Spark 1.2 on Inference360: place 45 of 283, index 64.9" height="20"></a>
Markdown
[![Muse Spark 1.2 on Inference360: place 45 of 283, index 64.9](https://inference360.ai/badges/muse-spark-1-2.svg)](https://inference360.ai/models/muse-spark-1-2)

Compare with

Muse Ember64.8GPT-5.464.8Muse Spark Open65.1Muse Spark 1.164.6GPT-5.6 Luna65.5

Lower price

dots3-note Preview63.4 · FreeGLM-5.3-Flash63.1 · $0.078/1MLing 3.1 Flash67.2 · $0.135/1M

More from Meta

All models
Muse Glimmer 30B53.0Muse Spark 1.2 ContributorUnrankedMuse Spark 1.164.6Muse Spark 1.370.1Muse Spark 1.3 ContributorUnranked