Muse Wonder Full
MetaProprietary
描述
Muse Wonder Full (muse-wonder-full-1.0) is Meta Superintelligence Labs' most capable Muse Wonder series model for personal and professional agentic tasks, coding, and computer use. It is available only in Muse Code on the Power Usage ($50/mo) and Maximum Usage ($200/mo) plans, and is not offered on the Meta Model API. Catalog card only: no ZeroEval playground routing or first-party API pricing row.
發布日期
2026-10-12
參數規模
—
上下文長度
—
支援模態
—
能力雷達圖
60
general
60
coding
50
reasoning
51
science估算
50
agents
50
multimodal
缺少專門科學評測時，Science 由 LLM Stats 科學得分或推理能力估算。
排行榜排名
基準測試分數 (LLM Stats)
(LLM Stats (zeroeval))Agents
MemoryBench
85.5%自報
JobBench
75.3%自報
Personal Superintelligence Index
59.1%自報
Program Bench
59.0%自報
FocusBench
54.2%自報
AutomationBench-AA
45.4%自報
Code
DeepSWE 1.1
79.0%自報
SWE-Marathon
57.5%自報
Terminal-Bench 4.0-AA
57.1%自報
Finance
Vals Finance Agent v2
59.1%自報
General
Artificial Analysis
56.0%自報
AA Intelligence Index v4.3
56.0%自報
GDP.pdf
33.2%自報
Healthcare
HealthBench Professional
62.3%自報
Knowledge
Harvey LAB-AA
13.1%自報
Multimodal
OSWorld 2.1 (partial)
78.4%自報
OSWorld 2.1 (strict)
39.4%自報
Reasoning
Terminal-Bench 4.0
57.1%自報
AA 評測指數
(Artificial Analysis)暫無 AA 評測資料
LLM Stats 分類評分
(LLM Stats (zeroeval))Long Context90
Productivity80
Finance60
General60
Healthcare60
Agents60
Code60
Knowledge50
Multimodal50
Reasoning50
Tool Calling50
Vision50
定價
暫無定價資料
速度
暫無速度資料
供應商價格排行
暫無提供商資料