Inference
EN

Muse Wonder Full vs Step 5 Preview

Change models
Muse Wonder FullMeta
Step 5 PreviewStepFun
IndexAll categories, out of 100.
77.3 ±5.3 best
68.4 ±3.7
GapIndex points below the top.
Top
−8.9
CoverageIndex weight that has results.
80%
90% best
SpeedOutput tokens per second.
—
105/s
EloLMArena rating and rank.
—
1443
ReleasedMonth of release.
2026-10
2026-09
Input / 1MUS dollars per 1M input tokens.
—
$1.00
Output / 1MUS dollars per 1M output tokens.
—
$2.70
1K chats2K tokens in, 500 out, each.
—
$3.35
1K documents20K tokens in, 1K out, each.
—
$22.70
1K agent tasks200K tokens in, 20K out, each.
—
$254.00
ContextMaximum tokens in one request.
1M
1M
AgenticMulti-step tasks with tools.
78.2 best
68.8
CodingCode writing and repair.
77.3 best
71.4
ReasoningLogic problems and puzzles.
75.6 best
67.4
MultimodalTasks with images and text.
—
62.2
KnowledgeFacts and expert knowledge.
71.7 best
63.5
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
—
MathMath problems.
79.6 best
61.7

Benchmarks

27 shared
Only Muse Wonder Full: 15Only Step 5 Preview: 35
Muse Wonder Full
Step 5 Preview
Artificial Analysis Humanity's Last ExamKnowledge
53.6% best9 / 222
46.5%23 / 222
Artificial Analysis Long Context ReasoningReasoning
83.7%10 / 217
88.3% best2 / 217
Critical Physics TasksReasoning
32.9% best2 / 214
20.9%26 / 214
Artificial Analysis Omniscience AccuracyKnowledge
41.0% best46 / 164
36.4%64 / 164
GDPval-AA normalizedAgentic
67.9% best2 / 133
54.3%24 / 133
Vibe Code Bench v1.1Coding
86.0% best14 / 123
70.4%48 / 123
Artificial Analysis SciCodeCoding
62.6% best4 / 114
58.9%13 / 114
Legal Research BenchLegal
53.4% best6 / 90
40.4%35 / 90
Tax Agent BenchFinance
43.0% best8 / 82
27.3%39 / 82
Finance Agent (v2)Finance
59.1% best6 / 79
50.7%42 / 79
Code MigrationCoding
61.0% best7 / 77
39.0%31 / 77
EMBFinance
71.9% best10 / 75
57.1%41 / 75
Terminal-Bench 4.0Agentic
58.1% best6 / 61
31.8%23 / 61
DeepSWEAgentic
79.0% best1 / 59
67.7%27 / 59
Agent Arena steerabilityAgentic
9.6 best12 / 54
−0.930 / 54
Agent Arena task outcomeAgentic
11.1 best9 / 54
−1.535 / 54
ProofBench v1.1Math
94.0% best11 / 54
42.0%37 / 54
Public Benefits Bench v1.1Public benefits
73.7% best3 / 50
58.1%37 / 50
IOICoding
79.8% best11 / 45
38.1%38 / 45
Terminal-Bench-ScienceScience
37.1% best8 / 44
2.9%33 / 44