Inference
EN
Meta

Muse Wonder Full

Muse Wonder Full is a reasoning model from Meta in the Muse Wonder family. 24 benchmarks count toward its index, in 5 categories.

  • Closed weights
  • Reasoning
  • Input: text
  • Released 2026-10
RankedEnough results for an index.
IndexAll categories, out of 100.77.3±5.3Median 51.0
CoverageIndex weight that has results.80%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.1MMedian 262K
EloLMArena rating and rank.—Median 1422

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.10 counted78.24 / 218
CodingCode writing and repair.7 counted77.36 / 209
ReasoningLogic problems and puzzles.3 counted75.615 / 227
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.3 counted71.713 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.None counted——
MathMath problems.1 counted79.6—

Results

24 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
Critical Physics TasksReasoningAbove fitFar above this model's index.32.9%2 / 214
GDPval-AA normalizedAgentic67.9%2 / 133
DeepSWEAgentic79.0%1 / 59
JobBenchAgentic75.3%1 / 36
Artificial Analysis SciCodeCoding62.6%4 / 114
Vibe Code Bench 1-100Coding34.0% ±5.31 / 26
Artificial Analysis Humanity's Last ExamKnowledge53.6%9 / 222
MazeBench: visual spatial reasoning in a 3D open worldAgentic17.0%2 / 48
Artificial Analysis GDP.pdfAgentic32.6%1 / 22
Artificial Analysis Long Context ReasoningReasoning83.7%10 / 217
Code MigrationCoding61.0% ±4.47 / 77
Terminal-Bench 4.0Agentic58.1% ±3.36 / 61
Vibe Code Bench v1.1Coding86.0% ±1.914 / 123
Agent Arena task outcomeAgentic11.19 / 54
ProofBench v1.1Math94.0% ±2.411 / 54
Agent Arena steerabilityAgentic9.612 / 54
Bug Hunt BenchCoding42.0%6 / 25
IOICoding79.8% ±3.011 / 45
Artificial Analysis Omniscience AccuracyKnowledge41.0%46 / 164
Artificial Analysis Terminal-Bench v4.0Agentic57.1%7 / 22
HealthBench ProfessionalKnowledge62.3%7 / 18
ProgramBench: Can Language Models Rebuild Programs From Scratch?Coding59.0%9 / 19
OSWorld 2.1 (all 108 tasks)Agentic78.4%6 / 11
Medical Long Context Reasoning (MLCR-AA)Reasoning22.2%12 / 22
Artificial Analysis AutomationBenchAgentic45.4%21 / 23
LegalBenchLegal87.4% ±0.44 / 149
Public Benefits Bench v1.1Public benefits73.7% ±1.13 / 50
Legal Research BenchLegal53.4% ±3.56 / 90
MedScribeHealthcare89.3% ±2.08 / 109
Finance Agent (v2)Finance59.1% ±0.26 / 79
Tax Agent BenchFinance43.0% ±3.68 / 82
EMBFinance71.9% ±2.710 / 75
HLAB-Vals-1.1Legal23.3% ±3.55 / 35
MedCodeHealthcare49.9% ±2.218 / 106
Terminal-Bench-ScienceScience37.1% ±5.88 / 44
CyberBench v1.1Other73.6% ±5.611 / 48
ProgramBenchCoding0.5% ±0.519 / 65
BioMysteryBenchScience73.3% ±0.69 / 30
SRE BenchOther11.8% ±2.012 / 36
MysteryMechanismScience35.1% ±3.211 / 30
SWE-MarathonCoding57.5%—
Harvey LAB-AA v1.1Agentic13.1%—

Sources

BenchLMCC BY-NC 4.0Vals AINo licence statedLMArenaCC BY 4.0

History

5 releases
Index77.7 → 77.3
23 Oct 202612 Nov 2026

Questions

How good is Muse Wonder Full?
Its index is 77.3 out of 100. That is place 7 of 283 ranked models.
What is Muse Wonder Full best at?
Agentic: 78.2 out of 100, place 4 of 218.
What is the context window of Muse Wonder Full?
1M tokens.
Is Muse Wonder Full open source?
No. Its weights are not published.
When was Muse Wonder Full released?
October 2026.

Badge

Muse Wonder Full on Inference360: place 7 of 283, index 77.3
HTML
<a href="https://inference360.ai/models/muse-wonder-full"><img src="https://inference360.ai/badges/muse-wonder-full.svg" alt="Muse Wonder Full on Inference360: place 7 of 283, index 77.3" height="20"></a>
Markdown
[![Muse Wonder Full on Inference360: place 7 of 283, index 77.3](https://inference360.ai/badges/muse-wonder-full.svg)](https://inference360.ai/models/muse-wonder-full)

Compare with

Claude Sonnet 5.577.2GPT-6.1 Terra77.1GPT-6.1 Sol78.0GPT-6 Astra78.0Claude Fable 5.176.2

More from Meta

All models
Muse Wonder73.1Muse Spark Open65.1Muse Ember64.8Muse Spark 1.3 ContributorUnrankedMuse Spark 1.370.1