跳转到主要内容

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

领域#排名分数来源
智能体能力模型榜67
54.0
LS
代码能力榜33
79.0
AA
通用能力榜19
81.0
AA
多模态榜77
60.0
LS
推理能力92
50.0
LS
科学能力15
84.0
AA

基准测试分数 (LLM Stats)

Agents

GDPval-AA1164.00 / 3000自报
DeepSearchQA74.8%自报
Terminal-Bench 2.059.0%自报
SWE-Bench Pro52.4%自报

Biology

GPQA89.5%自报

Code

LiveCodeBench Pro0.80 / 3000自报
SWE-Bench Verified77.4%自报

Communication

Tau2 Telecom91.5%自报

General

MMMU-Pro80.4%自报
SimpleVQA0.71 / 100自报

Grounding

ScreenSpot Pro84.1%自报

Healthcare

MedXpertQA78.4%自报
HealthBench Hard42.8%自报

Math

Humanity's Last Exam58.4%自报

Multimodal

CharXiv-R86.4%自报
ZEROBench0.33 / 100自报

Physics

IPhO 202582.6%自报

Reasoning

ERQA64.7%自报
ARC-AGI v242.5%自报
FrontierScience Research38.3%自报

AA 评测指数

Coding Index
58.6
Intelligence Index
43.1
Tau2
0.9
Gpqa
0.9
Ifbench
0.8
Lcr
0.7
Terminalbench V2 1
0.6
Scicode
0.5
Terminalbench Hard
0.5
Hle
0.4
Tau Banking
0.2

LLM Stats 分类评分

Legal
100
Finance
100
Agents
100
General
100
Reasoning
78
Physics
90
Biology
90
Chemistry
90
Communication
90
Frontend Development
80
Grounding
80
Tool Calling
80
Image To Text
70
Multimodal
70
Search
70
Code
70
Vision
70
Math
60
Spatial Reasoning
60
Healthcare
60

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

外部链接