Kimi K3.1 Pro Together vs Muse Spark 1.3
完整资料:Kimi K3.1 Pro Together(Moonshot AI)和 Muse Spark 1.3(Meta),包括全部基准测试、价格和排名。
更新于 · 数据来自 7 个来源
截至 2026年11月16日,Kimi K3.1 Pro Together 与 Muse Spark 1.3 的质量不相上下(BenchLeader Index 60.0 比 61.6,在 ±8.5 的误差范围内)。看重数学(65 比 56)选 Kimi K3.1 Pro Together;看重知识(81 比 66)和推理(71 比 67)选 Muse Spark 1.3。
要点(截至 2026年11月16日)
- Kimi K3.1 Pro Together 的 BenchLeader Index 为 60.0,Muse Spark 1.3 为 61.6(BenchLeader,2026年11月16日)。
- Muse Spark 1.3 每百万 token 混合价格为 $2.00(标价,2026年11月16日)。
按用途怎么选
推荐模型的数值列在前面。类别得分采用 BenchLeader 标尺(50 为模型平均水平,15 分为一个标准差)。领先半个标准差及以上为“高”置信度;相差不到 1 分则视为“均可”。
主要差异
- Kimi K3.1 Pro Together (max) 与 Muse Spark 1.3 的质量不相上下(60.0 比 61.6,在 ±8.5 的误差范围内)。
- Kimi K3.1 Pro Together (max) 在数学方面更强。
- Muse Spark 1.3 在知识、推理方面更强。
概览
| 规格 | Kimi K3.1 Pro Together | Muse Spark 1.3 |
|---|---|---|
| 提供商 | Moonshot AI | Meta |
| 发布日期 | 2026年10月23日 | 2026年9月2日 |
| 上下文窗口(token) | – | 1,048,576 |
| 对比时的配置 | Kimi K3.1 Pro Together (max) | Muse Spark 1.3 |
| 开放权重 | 否 | 否 |
| 每百万 token 价格(输入 / 输出 / 缓存输入) | – | $1.25 / $4.25 / $0.150 |
逐类别对比
类别得分采用 BenchLeader 标尺:50 为参评模型的平均水平,15 分为一个标准差。差距不到 1 分视为持平。
| 类别 | Kimi K3.1 Pro Together (max) | Muse Spark 1.3 | 领先者 |
|---|---|---|---|
| 推理 | 66.9 | 71.0 | Muse Spark 1.3 |
| 智能体与工具 | – | 61.3 | 无法比较 |
| 数学 | 65.2 | 56.3 | Kimi K3.1 Pro Together |
| 知识 | 66.0 | 81.2 | Muse Spark 1.3 |
同一图表对比
各领域采用与 Index 相同的标尺,平均模型为 50 分。单击某个领域可查看构成它的基准测试。 可从列表中最多选择 5 个模型。
未绘制(测得的模型少于两个):编程, 智能体与工具, 指令遵循, 长上下文, 多模态, 人类偏好。
各来源的分歧
在两者都有结果的 0 项基准测试中,Kimi K3.1 Pro Together 领先 0 项,Muse Spark 1.3 领先 0 项。
| 综合 Index | Kimi K3.1 Pro Together (max) | Muse Spark 1.3 | 领先者 |
|---|---|---|---|
| BenchLeader Index | 60.0 | 61.6 | 不相上下 |
各发布方使用各自的任务、评测框架和推理强度设置,因此总体结论可能不同;BenchLeader 将它们统一到同一标尺上。† 表示同一模型在另一推理强度设置下的测量值(发布方未测试本页对比的设置)。
价格、速度与延迟
标价按每 1 个输出 token 对应 3 个输入 token 混合计算。每项任务的价格采用方法论页面中的 token 数量。每项基准测试任务的成本、token 数和耗时均为实测值,取自 Artificial Analysis 整个 Intelligence Index 测试的平均值。
| 指标 | Kimi K3.1 Pro Together (max) | Muse Spark 1.3 |
|---|---|---|
| 每百万 token 混合价格 | – | $2.00 |
| 每百万 token 输入 / 输出 | – / – | $1.25 / $4.25 |
| 每百万 token 缓存输入 | – | $0.150 |
| 每项任务价格:聊天回复 | – | $0.0018 |
| 每项任务价格:总结一份 30 页的报告 | – | $0.018 |
| 每项任务价格:代码编辑 | – | $0.014 |
| 每项任务价格:智能体编程会话 | – | $0.092 |
| 每项任务价格:结构化信息提取 | – | $0.0034 |
| 智能体编程会话(输入大多命中缓存) | – | $0.043 |
| 输出速度 | – | 163 tok/s |
| 首次回答时间 | – | 2.6 秒 |
每个模型都按其排名最高的推理强度设置进行对比。Kimi K3.1 Pro Together (max) 得分 60.0。Muse Spark 1.3 得分 61.6,首个回答 token 需 3 秒;Muse Spark 1.3 (max) 为 60.6(31 秒),Muse Spark 1.3 (xhigh) 为 58.1(11 秒)。推理强度越高,回答前思考越久:得分更高,但等待更长、计费 token 也更多。 推理强度如何影响得分和成本
全部基准测试
共 123 项基准测试,按类别分组;点击名称可查看该排行榜及领先模型。† 表示同一模型在另一推理强度设置下的测量值,仅供参考,不计入胜场。
| 指标 | 发布方 | Kimi K3.1 Pro Together (max) | Muse Spark 1.3 |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 60.0 | 61.6 |
| 知识得分 | BenchLeader | 66.0 | 81.2 |
| 数学得分 | BenchLeader | 65.2 | 56.3 |
| 推理得分 | BenchLeader | 66.9 | 71.0 |
| 智能体与工具得分 | BenchLeader | – | 61.3 |
| 混合价格(美元/百万 token) | BenchLeader | – | $2.00 |
| 输出速度 | BenchLeader | – | 163 tok/s |
| 首次回答时间 | BenchLeader | – | 2.6 秒 |
| 上下文窗口 | BenchLeader | – | 1,048,576 token |
| 推理 | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 94.2% | – |
| SimpleBench | SimpleBench | – | 81.8% |
| Arena Hard Prompts | Arena (LMArena) | – | 1517† |
| LiveBench 2.0 Reasoning | LiveBench | – | 89.7%† |
| LiveBench 1.0 Reasoning | LiveBench | – | 89.7%† |
| GPQA Diamond (AA) | Artificial Analysis | – | 93.5%† |
| Humanity's Last Exam (AA) | Artificial Analysis | – | 48.7%† |
| HLE-Diamond (AA) | Artificial Analysis | – | 33.7%† |
| CritPt | Artificial Analysis | – | 24.9%† |
| CritPt v2 hard subset (AA) | Artificial Analysis | – | 48.0%† |
| MysteryMechanism | Vals AI | – | 36.0%† |
| HLE-Diamond (Scale) | Scale AI SEAL | – | 25.4% |
| Arena Expert | Arena (LMArena) | – | 1521† |
| Chess Puzzles | Epoch AI Benchmarking Hub | 45.0% | 37.0%† |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 42.0% | 25.0%† |
| LMCA | Epoch AI Benchmarking Hub | – | 53.9% |
| DTBench | Epoch AI Benchmarking Hub | – | 96.5% |
| ForecastBench | ForecastBench | – | 65.5%† |
| BTF-3 | Epoch AI Benchmarking Hub | – | 14.1%† |
| 编程 | |||
| SciCode | SciCode | – | 58.8%† |
| Arena Coding | Arena (LMArena) | – | 1534† |
| Arena WebDev | Arena (LMArena) | – | 1655† |
| LiveBench 2.0 Coding | LiveBench | – | 84.4%† |
| LiveBench 1.0 Coding | LiveBench | – | 81.1%† |
| IOI | Vals AI | – | 56.6%† |
显示全部 123 项基准测试
| 指标 | 发布方 | Kimi K3.1 Pro Together (max) | Muse Spark 1.3 |
|---|---|---|---|
| 编程 | |||
| Code Migration | Vals AI | – | 47.4%† |
| ProgramBench | Vals AI | – | 2.5%† |
| SWE-bench Multimodal | Vals AI | – | 43.1%† |
| Vibe Code Bench 1-100 | Vals AI | – | 20.5%† |
| Vibe Code Bench v1.1 | Vals AI | – | 85.9%† |
| MirrorCode | Epoch AI Benchmarking Hub | – | 54.4%† |
| CursorBench | Cursor | – | 41.6%† |
| ALE-Bench | Epoch AI Benchmarking Hub | – | 1688.3† |
| FrontierSWE | Epoch AI Benchmarking Hub | – | 25.8%† |
| WeirdML v3 | WeirdML | – | 6.7%† |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | – | 33.3%† |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | – | 84.3%† |
| Startrise Frontend | Startrise | – | 74.7% |
| GameBench 2 Overall | DuelLab GameBench 2 | – | 57.7 |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | – | 68.5† |
| Design Arena Website | Design Arena | – | 1353† |
| FrontierCode 1.1 (AA) | Artificial Analysis | – | 40.6%† |
| FrontierCode 2.0 (AA) | Artificial Analysis | – | 23.1%† |
| 智能体与工具 | |||
| Terminal-Bench | Terminal-Bench | – | 14.8%† |
| APEX-Agents | Mercor | – | 58.6% |
| Arena Agent | Arena (LMArena) | – | 2.2† |
| LiveBench Long Agentic | LiveBench | – | 47.9%† |
| LiveBench Browser | LiveBench | – | 30.9%† |
| LiveBench Agentic Coding | LiveBench | – | 64.1%† |
| Terminal-Bench 2.1 (Vals) | Vals AI | – | 79.0%† |
| HiL-Bench | Scale AI SEAL | – | 66.9% |
| GDPval-AA v2.1 | Artificial Analysis | – | 1681† |
| τ³-Banking (AA) | Artificial Analysis | – | 50.5%† |
| ITBench SRE (AA) | Artificial Analysis | – | 33.2%† |
| Analyst Agent (AA) | Artificial Analysis | – | 37.5%† |
| CUA-bench | Vals AI | – | 5.8%† |
| CyberBench | Vals AI | – | 72.7%† |
| Finance Agent v2 | Vals AI | – | 60.0%† |
| Harvey's Legal Agent Benchmark v1 | Vals AI | – | 23.8%† |
| Legal Research Bench | Vals AI | – | 55.3%† |
| SREBench | Vals AI | – | 3.8%† |
| Tax Agent Bench | Vals AI | – | 42.4%† |
| Terminal-Bench 4.0 (Vals) | Vals AI | – | 24.8%† |
| Terminal-Bench Science | Vals AI | – | 10.0%† |
| Arena Agent Task Completion | Arena (LMArena) | – | 3.7† |
| AutomationBench | Artificial Analysis | – | 57.9%† |
| GDP.pdf (AA) | Artificial Analysis | – | 26.6%† |
| Harvey LAB-AA v1.1 | Harvey | – | 8.9%† |
| AA-Briefcase v1.1 | Artificial Analysis | – | 1581† |
| Design Arena Agentic | Design Arena | – | 1283† |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | – | 10.9%† |
| CWE-Bench (AA) | Artificial Analysis | – | 60.8%† |
| DeepSec-Bench (AA) | Artificial Analysis | – | 12.5%† |
| CyberGym end-to-end (AA) | Artificial Analysis | – | 59.5%† |
| 数学 | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 78.3% | 74.0%† |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 65.8% | 46.3%† |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 100.0% | 99.2%† |
| ProofBench | Vals AI | – | 55.0% |
| LiveBench Mathematics | LiveBench | – | 96.0%† |
| Arena Maths | Arena (LMArena) | – | 1506† |
| 知识 | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 62.2% | – |
| LiveBench 2.0 Data Analysis | LiveBench | – | 79.6%† |
| LiveBench Research | LiveBench | – | 82.5%† |
| LiveBench 1.0 Data Analysis | LiveBench | – | 79.6%† |
| AA-Omniscience v1.1 | Artificial Analysis | – | 22.4† |
| PRBench Finance | Scale AI SEAL | – | 59.5% |
| PRBench Legal | Scale AI SEAL | – | 61.6% |
| Excel Modeling Benchmark | Vals AI | – | 67.4%† |
| Access to Justice Benchmark | Vals AI | – | 72.2%† |
| Financial Coaching Bench | Vals AI | – | 70.3%† |
| Immigration Bench | Vals AI | – | 64.9%† |
| DrugDiscoveryBench | Scale AI SEAL | – | 62.2%† |
| GDP.pdf | Epoch AI Benchmarking Hub | – | 27.6% |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | – | 38.0%† |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | – | 74.8%† |
| 指令遵循 | |||
| LiveBench Language | LiveBench | – | 82.8%† |
| LiveBench Instruction Following | LiveBench | – | 78.0%† |
| Arena Instruction Following | Arena (LMArena) | – | 1485† |
| 人类偏好 | |||
| Arena Text | Arena (LMArena) | – | 1497† |
| EQ-Bench Creative Writing v3 | EQ-Bench | – | 1907 |
| Arena Creative Writing | Arena (LMArena) | – | 1468† |
| Arena Multi-turn | Arena (LMArena) | – | 1493† |
| Arena Longer Queries | Arena (LMArena) | – | 1498† |
| 多模态 | |||
| Arena Vision | Arena (LMArena) | – | 1310† |
| LiveBench Design | LiveBench | – | 74.8%† |
| MMMU-Pro | Artificial Analysis | – | 82.0%† |
| Humanity's Sixth Sense | Scale AI SEAL | – | 37.4%† |
| GeoBench | GeoBench | – | 86.0% |
| Furniture Assembly | Epoch AI Benchmarking Hub | 55.8% | – |
| Code2Video-AA | Artificial Analysis | – | 1019† |
| AA-3D | Artificial Analysis | – | 1032† |
| 长上下文 | |||
| AA-LCR v1.1 | Artificial Analysis | – | 83.0%† |
| Arena Document | Arena (LMArena) | – | 1474† |
| MLCR | Artificial Analysis | – | 13.9%† |
| 综合 | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | – | 158.2 |
| LiveBench 2.0 | LiveBench | – | 68.6%† |
| LiveBench 1.0 | LiveBench | – | 81.6%† |
| AA Intelligence Index v5.0.1 | Artificial Analysis | – | 36.6† |
| AA Intelligence Index v4.3.2 | Artificial Analysis | – | 48.1† |
| AA Progress Index v1 | Artificial Analysis | – | 324.5† |
| Vals RSI Index | Vals AI | – | 19.6%† |
| AA Cyber Index | Artificial Analysis | – | 44.3† |
| 安全与诚实 | |||
| DistressBench | Scale AI SEAL | – | 84.9% |
数据截至 2026年11月16日。对比的是每个模型的最佳配置;每项分数的来源链接都在模型页面上。
DuelLab GameBench 2, release gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
变化
- Muse Spark 1.3:BenchLeader Index 60.2 → 61.6。
- Muse Spark 1.3:1 项分数修订。
- Kimi K3.1 Pro Together (max):BenchLeader Index 60.5 → 60.0。
- Muse Spark 1.3:BenchLeader Index 59.2 → 60.2。
- Muse Spark 1.3:2 项新结果(SimpleBench, GeoBench)。
- Muse Spark 1.3:BenchLeader Index 57.7 → 59.2。
- Muse Spark 1.3:4 项分数修订。
- Kimi K3.1 Pro Together (max):首批结果。
- Muse Spark 1.3:BenchLeader Index 58.2 → 57.7。
- Muse Spark 1.3:1 项分数修订。
相关对比
- Kimi K3.1 Pro Together vs Claude Fable 5.2
- Kimi K3.1 Pro Together vs Gemini 4 Carbon
- Kimi K3.1 Pro Together vs Claude Opus 5.5
- Muse Spark 1.3 vs Claude Fable 5.2
- Muse Spark 1.3 vs Gemini 4 Carbon
- Muse Spark 1.3 vs Claude Opus 5.5
- Muse Spark 1.3 vs GPT-6 Astra
- Kimi K3.1 Pro Together vs Claude Fable 5.1
排行榜: 综合最佳大模型 · 最适合智能体的大模型 · 便宜好用的大模型
Kimi K3.1 Pro Together vs Muse Spark 1.3:常见问题
- Kimi K3.1 Pro Together 比 Muse Spark 1.3 更好吗?
- Kimi K3.1 Pro Together (max) 与 Muse Spark 1.3 的质量不相上下(60.0 比 61.6,在 ±8.5 的误差范围内)。BenchLeader Index 综合了所有第三方质量基准测试;截至 2026年11月16日,两者差距在误差范围内,请结合你的用途查看各类别得分。
- Kimi K3.1 Pro Together 是什么时候发布的?
- Kimi K3.1 Pro Together 于 2026年10月23日 发布。