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.
发布日期
2026-04-08
参数规模
—
上下文长度
—
支持模态
—
能力雷达图
42
general
58
coding
88
reasoning
66
science估算
80
agents
70
multimodal
Science 在缺少专门科学评测时使用推理能力代理估算。
排行榜排名
基准测试分数 (LLM Stats)
Agents
GDPval-AA
1164.00 / 3000自报
DeepSearchQA
74.8%自报
Terminal-Bench 2.0
59.0%自报
SWE-Bench Pro
52.4%自报
Biology
GPQA
89.5%自报
Code
LiveCodeBench Pro
0.80 / 3000自报
SWE-Bench Verified
77.4%自报
Communication
Tau2 Telecom
91.5%自报
General
MMMU-Pro
80.4%自报
SimpleVQA
0.71 / 100自报
Grounding
ScreenSpot Pro
84.1%自报
Healthcare
MedXpertQA
78.4%自报
HealthBench Hard
42.8%自报
Math
Humanity's Last Exam
58.4%自报
Multimodal
CharXiv-R
86.4%自报
ZEROBench
0.33 / 100自报
Physics
IPhO 2025
82.6%自报
Reasoning
ERQA
64.7%自报
ARC-AGI v2
42.5%自报
FrontierScience Research
38.3%自报
AA 评测指数
Coding Index58.6
Intelligence Index43.1
Tau20.9
Gpqa0.9
Ifbench0.8
Lcr0.7
Terminalbench V2 10.6
Scicode0.5
Terminalbench Hard0.5
Hle0.4
Tau Banking0.2
LLM Stats 分类评分
Legal100
Finance100
Agents100
General100
Reasoning78
Physics90
Biology90
Chemistry90
Communication90
Frontend Development80
Grounding80
Tool Calling80
Image To Text70
Multimodal70
Search70
Code70
Vision70
Math60
Spatial Reasoning60
Healthcare60
定价
输入价格免费
输出价格免费
混合价格(3:1)免费
速度
Tokens/秒0.0
首Token延迟0.00s
首回答延迟0.00s
供应商价格排行
暂无提供商数据