Muse Spark
MetaProprietary
描述
Muse Spark is the first model in the Muse family developed by Meta Superintelligence Labs. It is a natively multimodal reasoning model with support for tool-use, visual chain of thought, and multi-agent orchestration. It features a Contemplating mode that orchestrates multiple agents reasoning in parallel. It demonstrates competitive performance in multimodal perception, reasoning, health, and agentic tasks, with Contemplating mode achieving 58% on Humanity's Last Exam and 38% on FrontierScience Research.
發布日期
2026-04-08
參數規模
—
上下文長度
—
支援模態
—
能力雷達圖
49
general
48
coding
88
reasoning
66
science估算
80
agents
70
multimodal
Science 在缺少專門科學評測時使用推理能力代理估算。
排行榜排名
基準測試分數 (LLM Stats)
Agents
GDPval-AA
1444.00 / 3000自報
DeepSearchQA
74.8%自報
Terminal-Bench 2.0
59.0%自報
SWE-Bench Pro
52.4%自報
Biology
GPQA
89.5%自報
Code
LiveCodeBench Pro
0.80 / 3000自報
SWE-Bench Verified
77.4%自報
Communication
Tau2 Telecom
91.5%自報
General
MMMU-Pro
80.4%自報
SimpleVQA
0.71 / 100自報
Grounding
ScreenSpot Pro
84.1%自報
Healthcare
MedXpertQA
78.4%自報
HealthBench Hard
42.8%自報
Math
Humanity's Last Exam
58.4%自報
Multimodal
CharXiv-R
86.4%自報
ZEROBench
0.33 / 100自報
Physics
IPhO 2025
82.6%自報
Reasoning
ERQA
64.7%自報
ARC-AGI v2
42.5%自報
FrontierScience Research
38.3%自報
AA 評測指數
Intelligence Index52.1
Coding Index47.5
Tau20.9
Gpqa0.9
Ifbench0.8
Lcr0.7
Scicode0.5
Terminalbench Hard0.5
Hle0.4
LLM Stats 分類評分
Finance100
Legal100
Agents100
General100
Reasoning97
Biology90
Chemistry90
Communication90
Physics90
Tool Calling80
Frontend Development80
Grounding80
Vision70
Code70
Image To Text70
Multimodal70
Search70
Spatial Reasoning60
Healthcare60
Math60
定價
輸入價格免費
輸出價格免費
混合價格(3:1)免費
速度
Tokens/秒0.0 tokens/s
首Token延遲0.00s
首回答延遲0.00s
可用提供商
(LS 內部計價單位)暫無提供商資料