Muse Spark 1.3

Muse Spark 1.3 is a reasoning model from Meta in the Muse Spark family. 37 benchmarks count toward its score, in 6 categories.

availableShows if the model has enough results for an index.
IndexOverall score out of 100.75.0 ±4.2
CoverageShare of the index weight with results.85%
SpeedOutput tokens per second.66/s
Input / 1MUS dollars per 1M input tokens.$1.25
Output / 1MUS dollars per 1M output tokens.$4.25
ContextMaximum tokens in one request.1.05M
EloLMArena rating and rank.1490 (#10)

The index is a score out of 100. The ± range shows how much it can change.

4,723 votes. Elo shows what people prefer. It does not change the score.

CapabilitiesScore per category, out of 100.

Out of 100
AgenticMulti-step tasks with tools.
78.0
CodingCode writing and repair.
70.5
ReasoningLogic problems and puzzles.
73.1
MultimodalTasks with images and text.
N/A
KnowledgeFacts and expert knowledge.
72.5
MultilingualTasks in many languages.
N/A
InstructionTasks with strict rules in the prompt.
83.0
MathMath problems.
71.6

Results

37 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result as a score out of 100.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
OTIS Mock AIME 2024-2025Math99.2%67.0xhigh effortEpoch AI
OpenAI MRCR v2 8-needle 256K-512KReasoning98.5%Meta Superintelligence Labs
OpenAI MRCR v2 8-needle 512K-1MReasoning98.1%Meta Superintelligence Labs
LiveBench MathematicsMath95.9%76.1xhigh effort25 Jun 2026LiveBench
Artificial Analysis GPQA DiamondKnowledge93.5%64.4Artificial Analysis
LiveBench ReasoningReasoning89.7%80.5xhigh effort25 Jun 2026LiveBench
DeepSearchQAAgentic89.4%69.1Meta AI
Terminal-Bench 2.1 (provider run)Agentic88.8%76.5DeepSeek-AI
Terminal-Bench 2.1 (provider run)Agentic88.8%76.5DeepSeek-AI
Artificial Analysis Long Context ReasoningReasoning83.0%65.7Artificial Analysis
Vibe Code Bench v1.1Coding82.9%76.6OpenHands21 Sept 2026Vals AI
LiveBench LanguageKnowledge82.8%70.4xhigh effort25 Jun 2026LiveBench
LiveBench CodingCoding81.1%72.2xhigh effort25 Jun 2026LiveBench
LiveBench Data AnalysisReasoning79.6%66.5xhigh effort25 Jun 2026LiveBench
LiveBench Instruction FollowingInstruction78.0%83.0xhigh effort25 Jun 2026LiveBench
Artificial Analysis Coding IndexCoding75.8%72.4Artificial Analysis
DeepSWEAgentic75.4%78.6Datacurve AI
FrontierMath-Tiers-1-3-v2-PrivateMath74.0%73.1max effortEpoch AI
Terminal-Bench 2.1Agentic72.3%66.721 Sept 2026Vals AI
OSWorld 2.0Agentic66.9%87.8Mengqi Yuan et al.
JobBenchAgentic64.9%78.7Yuetai Li et al.
LiveBench Agentic CodingAgentic64.1%77.0xhigh effort25 Jun 2026LiveBench
SWE-Atlas Codebase QnACoding59.4%Meta Superintelligence Labs and Scale AI
Artificial Analysis SciCodeCoding58.8%74.5Artificial Analysis
GDPval-AA normalizedAgentic58.7%80.6Artificial Analysis
Artificial Analysis AutomationBenchAgentic57.9%69.2Artificial Analysis
Artificial Analysis Agentic IndexAgentic55.7%82.6Artificial Analysis
Artificial Analysis Tau3-BankingAgentic50.5%81.2Artificial Analysis
AutomationBenchAgentic49.4%95.0Moonshot AI
Artificial Analysis Humanity's Last ExamKnowledge48.7%77.4Artificial Analysis
Artificial Analysis Intelligence IndexKnowledge48.1%82.5Artificial Analysis
FrontierMath-Tier-4-v2-PrivateMath46.3%70.2max effortEpoch AI
IOICoding43.9%64.821 Sept 2026Vals AI
Artificial Analysis Omniscience AccuracyKnowledge43.6%67.8Artificial Analysis
Medical Long Context Reasoning (MLCR-AA)Reasoning43.3%75.6Wisedocs and Artificial Analysis
Chess PuzzlesReasoning38.0%72.6max effortEpoch AI
Code MigrationCoding27.6%62.621 Sept 2026Vals AI
Artificial Analysis GDP.pdfAgentic26.6%78.1Artificial Analysis
Mystery Game PuzzlesReasoning25.0%60.5max effortEpoch AI
Critical Physics TasksReasoning24.9%90.6Artificial Analysis
ApprenticeBench: end-to-end computer use, continual learning, and long-horizon agency on a real accounts-payable jobAgentic19.0%70.7NeoCognition
Terminal-Bench 4.0Agentic15.2%70.121 Sept 2026Vals AI
Agent Arena task outcomeAgentic10.379.0max effort15 Sept 2026LMArena
Agent Arena command recoveryAgentic5.974.0max effort15 Sept 2026LMArena
Agent Arena steerabilityAgentic1.569.0max effort15 Sept 2026LMArena

37 benchmarks count, from 42 of 45 results. A grey row does not count. Too few models took that benchmark.

Sources

BenchLM benchmark aggregationUsed with attribution; per-benchmark results credited to their original authorsOpenRouter, collected directlyNo licence statedEpoch AI, collected directlyCC BY — free to use and redistribute with attributionLiveBench, collected directlyNo licence stated for the leaderboard. The site repo serving these CSVs has no LICENSE; the harness repo carries upstream Apache-2.0 and MIT copies that cover code, not resultsVals AI, collected directlyNo licence stated. Read from the public leaderboard and credited to Vals AILMArena, collected directlyCC BY 4.0 (lmarena-ai/leaderboard-dataset on Hugging Face)

Same level, lower price

MiMo-V2.6-Pro73.5 · $0.87Step 5 Preview72.5 · $2.7Gemini 3.7 Flash72.3 · $3.75

More from Meta

Muse Spark 1.269.2Muse Spark 1.167.0Muse Spark61.6Muse Glimmer 30B53.5Llama 4 Maverick33.6