Muse Spark 1.1 is a multimodal reasoning model from Meta, built for agentic tasks. It accepts text, images, video, and PDF documents and returns text, with a 1M-token context window.
The model is designed to orchestrate multi-agent workflows, acting as either a main agent that plans and delegates or as a subagent, and generalizes zero-shot to new tools, MCP servers, and custom skills. It supports structured output, parallel function calling, built-in search with citations, and configurable reasoning effort. Meta reports strong performance on real-world coding across large codebases, computer-use workflows, and visual-to-code generation.
提供商
此模型由一家提供商托管。OpenRouter 直接转发每个请求,不进行路由。
| US$1.25 | US$4.25 | US$0.15 | 1.69s | 132 tok/s |
- 输入 /百万
- US$1.25
- 输出 /百万
- US$4.25
- 缓存读取 /百万
- US$0.15
- 延迟
- 1.69s
- 吞吐量
- 132 tok/s
- 正常运行时间
- 100.00%
定价
客户为此模型实际支付的平均价格,与提供商公布的价格并列展示。由于缓存和折扣,实际支付的价格往往远低于标价。
加权平均值
Muse Spark 1.1 — 价格历史
| US$0.448 | US$4.25 | 72.9% | 100.0% |
- 有效输入价格 US$0.448 / 实际输出 US$4.25token 占比100.0%
性能
吞吐量是模型生成内容的速度(每秒 token 数,越高越好)。延迟是完整的往返时间(越低越好)。TTFT 是首个 token 的响应时间,即您看到内容出现前需要等待的时间(越低越好)。
P50,所有提供商中最佳
P50,最佳提供商
正常运行时间
正常运行时间是指过去 3 天中至少有一家提供商响应请求的时间百分比。可用性是指成功提供推理服务的时间百分比。OpenRouter 会持续监控,并在某家提供商返回错误时使用次优提供商。
过去 3 天的可用性
最近 72 小时过去 24 小时的可用性
当上游提供商发生错误时,如果您的请求筛选条件允许,我们可以通过路由到另一家健康的提供商来恢复服务。您可以通过以下接口以编程方式访问各提供商的正常运行时间数据: 端点 API。了解更多有关我们的 负载均衡和提供商路由选项。
基准测试
标准化评测得分,以及此模型针对共用基准测试提示词的实际输出。百分比越高越好,每项排名显示此模型在该来源所评测的模型中的位置。
| 来源 | 基准测试 | 得分 |
|---|---|---|
| Artificial Analysis | Muse Spark 1.1 (Xhigh) AutomationBench-AA | 38.8% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) GDP.pdf | 14.4% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) GPQA Diamond | 89.8% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) HLE | 46.2% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) AA-LCR | 77.7% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) τ³-Banking | 31.8% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) GDPval-AA | 35.8% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) CritPt | 15.1% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) SciCode | 58.8% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) Terminal-Bench 4.0 | 6.1% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) Terminal-Bench 2.1 | 77.9% |
| Artificial Analysis | Muse Spark 1.1 (Xhigh) MLCR-AA | 7.2% |
| Design Arena | Muse Spark 1.1 Agents Arena Agenticgamedev Elo | 1130 |
| Design Arena | Muse Spark 1.1 Agents Arena Androidnative Elo | 1172 |
| Design Arena | Muse Spark 1.1 Agents Arena Blender3d Elo | 1009 |
| Design Arena | Muse Spark 1.1 Agents Arena Codetovideo Elo | 1091 |
| Design Arena | Muse Spark 1.1 Agents Arena Full Stack Elo | 1167 |
| Design Arena | Muse Spark 1.1 Agents Arena Godotgamedev Elo | 1081 |
| Design Arena | Muse Spark 1.1 Agents Arena Htmlslides Elo | 1094 |
| Design Arena | Muse Spark 1.1 Agents Arena Mobile Apps Elo | 1170 |
| Design Arena | Muse Spark 1.1 Agents Arena Pdf Elo | 1158 |
| Design Arena | Muse Spark 1.1 Agents Arena Python-Pptxslides Elo | 1124 |
| Design Arena | Muse Spark 1.1 Agents Arena Spreadsheet Elo | 1001 |
| Design Arena | Muse Spark 1.1 Agents Arena Webapps Elo | 1185 |
| Design Arena | Muse Spark 1.1 Models Arena 3D Elo | 1230 |
| Design Arena | Muse Spark 1.1 Models Arena Asciiart Elo | 1245 |
| Design Arena | Muse Spark 1.1 Models Arena Code Categories Elo | 1258 |
| Design Arena | Muse Spark 1.1 Models Arena Data Visualization Elo | 1262 |
| Design Arena | Muse Spark 1.1 Models Arena Game Development Elo | 1250 |
| Design Arena | Muse Spark 1.1 Models Arena SVG Elo | 1219 |
| Design Arena | Muse Spark 1.1 Models Arena UI Component Elo | 1277 |
| Design Arena | Muse Spark 1.1 Models Arena Website Elo | 1260 |
| Vals AI | Muse Spark 1.1 Code Migration | 31.1% |
| Vals AI | Muse Spark 1.1 CorpFin v2 | 71.3% |
| Vals AI | Muse Spark 1.1 EMB | 56.4% |
| Vals AI | Muse Spark 1.1 Finance Agent (v2) | 57.2% |
| Vals AI | Muse Spark 1.1 GPQA Diamond | 91.2% |
| Vals AI | Muse Spark 1.1 Harvey's Legal Agent Benchmark v1 | 20.0% |
| Vals AI | Muse Spark 1.1 LiveCodeBench | 85.9% |
| Vals AI | Muse Spark 1.1 LegalBench | 85.0% |
| Vals AI | Muse Spark 1.1 Legal Research Bench | 38.0% |
| Vals AI | Muse Spark 1.1 MedScribe | 88.9% |
| Vals AI | Muse Spark 1.1 MMLU Pro | 88.7% |
| Vals AI | Muse Spark 1.1 MMMU Pro | 86.6% |
| Vals AI | Muse Spark 1.1 MortgageTax | 66.1% |
| Vals AI | Muse Spark 1.1 ProgramBench | 0.0% |
| Vals AI | Muse Spark 1.1 ProofBench v1 | 39.0% |
| Vals AI | Muse Spark 1.1 Public Benefits Bench v1.1 | 61.3% |
| Vals AI | Muse Spark 1.1 SAGE v1 | 45.6% |
| Vals AI | Muse Spark 1.1 SkillsBench | 59.2% |
| Vals AI | Muse Spark 1.1 SWE-bench | 82.0% |
| Vals AI | Muse Spark 1.1 TaxEval v2 | 79.7% |
| Vals AI | Muse Spark 1.1 Terminal-Bench 2.1 | 69.3% |
| Vals AI | Muse Spark 1.1 Vals Index | 54.8% |
| Vals AI | Muse Spark 1.1 Vals Multimodal Index | 66.7% |
| Vals AI | Muse Spark 1.1 Vibe Code Bench v1.1 | 72.2% |
应用
向此模型发送最多流量的公开应用。这能很好地反映实际生产工作负载的情况,也能提示此模型最适合哪些用例。
活动
此模型的 token 用量和请求流量随时间的变化。
API
可直接用于调用此模型的代码。OpenRouter 的 API 与 OpenAI 兼容——大多数 SDK 只需替换基础 URL 即可使用。切换模型时,唯一需要更改的是下方的模型标识符。
目前无法通过 EU 区域内路由使用。 区域内路由文档
常见问题
Muse Spark 1.1 is a multimodal reasoning model from Meta, built for agentic tasks. It accepts text, images, video, and PDF documents and returns text, with a 1M-token context window. The model is designed to orchestrate multi-agent workflows, acting as either a main agent that plans and delegates or as a subagent, and generalizes zero-shot to new tools, MCP servers, and custom skills.
Muse Spark 1.1 的费用为 US$1.25/百万输入 token 和 US$4.25/百万输出 token,Cache Read,价格为 US$0.15/M tokens和Web Search,价格为 US$2.50/1K calls 单独计费。
Muse Spark 1.1 的上下文窗口为 1,048,576 token。
是的。Muse Spark 1.1 接受 tools 和 tool_choice,以进行函数调用。 它还支持通过 response_format 中的 JSON schema 实现结构化输出。
Muse Spark 1.1 接受 文本、图像、视频和PDF 等文件 作为输入,并返回 文本。
在 OpenRouter 上,Muse Spark 1.1 由 1 家提供商提供服务:Meta。
Muse Ember, Muse Spark Open, Muse Wonder 及另外 5 个 是 Meta 的其他 文本 模型。
Muse Spark 1.1 于 2026年7月16日 发布。
