Inference
EN

Hy4 vs Muse Wonder

Change models
Hy4Tencent
Muse WonderMeta
IndexAll categories, out of 100.
72.0 ±7.8
73.1 ±4.0 best
GapIndex points below the top.
−1.1 · no clear gap
Top
CoverageIndex weight that has results.
65%
85% best
SpeedOutput tokens per second.
52/s
80/s best
EloLMArena rating and rank.
—
1503
ReleasedMonth of release.
2026-11
2026-10
Input / 1MUS dollars per 1M input tokens.
$0.417 best
$4.00
Output / 1MUS dollars per 1M output tokens.
$1.25 best
$20.00
1K chats2K tokens in, 500 out, each.
$1.46 best
$18.00
1K documents20K tokens in, 1K out, each.
$9.59 best
$100.00
1K agent tasks200K tokens in, 20K out, each.
$108.41 best
$1,200.00
ContextMaximum tokens in one request.
1.05M
1.05M
AgenticMulti-step tasks with tools.
74.5 best
73.0
CodingCode writing and repair.
68.4
72.3 best
ReasoningLogic problems and puzzles.
—
73.1
MultimodalTasks with images and text.
57.4
—
KnowledgeFacts and expert knowledge.
83.7 best
68.6
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
74.4
MathMath problems.
—
73.5

Benchmarks

12 shared
Only Hy4: 9Only Muse Wonder: 50
Hy4
Muse Wonder
Vibe Code Bench v1.1Coding
81.0%30 / 123
83.1% best22 / 123
Legal Research BenchLegal
41.3%30 / 90
50.5% best7 / 90
Tax Agent BenchFinance
34.4%17 / 82
36.3% best15 / 82
Terminal-Bench 4.0Agentic
39.4%15 / 61
49.5% best11 / 61
DeepSWEAgentic
74.9%8 / 59
78.8% best2 / 59
Agent Arena steerabilityAgentic
8.8 best14 / 54
2.622 / 54
Agent Arena task outcomeAgentic
6.7 best12 / 54
5.413 / 54
HLAB-Vals-1.1Legal
12.5%17 / 35
23.8% best4 / 35
FrontierCode 2.0Coding
31.0% best9 / 27
24.3%15 / 27
ProgramBench: Can Language Models Rebuild Programs From Scratch?Coding
30.0%12 / 19
50.0% best11 / 19
OSWorld 2.1 (all 108 tasks)Agentic
78.9% best5 / 11
72.6%9 / 11
SWE-MarathonCoding
43.8%
53.1%