Meta发布Muse Spark Open开源模型，支持M5 Ultra本地部署
Muse Spark Open
重磅开源模型发布，且首次实现超大参数模型在消费级Apple Silicon上的高效本地运行，对端侧AI开发者和硬件极客极具参考价值。
Muse Spark Open is available today and brings leading agentic capabilities to your desktop for the first time on Apple M5 Ultra Mac Studios.
Muse Spark Open 今日上线，首次将领先的智能体能力带入 Apple M5 Ultra Mac Studio 桌面端。
Today, we’re releasing the weights for Muse Spark Open, the most capable US open-weight model to date. Muse Spark Open is a natively multimodal model with 816B total parameters (42B active) and a 1M-token context window. Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning.
今天，我们发布了 Muse Spark Open 的权重，这是迄今为止最强大的美国开源权重模型。Muse Spark Open 是一个原生多模态模型，拥有 816B 总参数（42B 活跃参数）和 1M-token 上下文窗口。专为编码和智能体工作构建，它在长周期任务、复杂软件工程和多步推理方面表现出色。
We believe the future is for everyone, and that includes the freedom to run advanced models wherever, including at home. Muse Spark Open is the first model in its class to run on your desktop using two Apple M5 Ultra Mac Studios with the latest MLX framework.
我们相信未来属于每个人，这也包括在任何地方运行先进模型的自由，包括在家中。Muse Spark Open 是首款使用最新 MLX 框架在配备两台 Apple M5 Ultra Mac Studio 的桌面上运行的同类模型。
Muse Spark Open is available today through API platforms at Baseten, Fireworks AI, Groq, Modal, Nebius, OpenRouter and Together AI. Model weights are available for download on Amazon SageMaker AI, Google Cloud Model Garden, Hugging Face, and Microsoft Foundry Models.
Muse Spark Open 今日可通过 Baseten、Fireworks AI、Groq、Modal、Nebius、OpenRouter 和 Together AI 等 API 平台获取。模型权重可在 Amazon SageMaker AI、Google Cloud Model Garden、Hugging Face 和 Microsoft Foundry Models 上下载。
It runs on leading inference engines like SGLang, vLLM, TensorRT-LLM, Hugging Face Transformers, Apple MLX, llama.cpp, and Unsloth with optimizations for AMD, Apple, Intel, and NVIDIA silicon. Developers can easily further customize the model using the PyTorch framework.
它支持 SGLang、vLLM、TensorRT-LLM、Hugging Face Transformers、Apple MLX、llama.cpp 和 Unsloth 等主流推理引擎，并针对 AMD、Apple、Intel 和 NVIDIA 芯片进行了优化。开发者可以使用 PyTorch 框架轻松进一步定制该模型。
Key Benchmarks
关键基准测试
For more details about our evaluations, see the evaluation methodology report.
有关我们评估的更多详细信息，请参阅评估方法论报告。
Optimized for Apple Silicon
针对 Apple Silicon 优化
Leveraging the latest version of MLX, Muse Spark Open was optimized to run on Mac Studio with M5 Ultra, and together they’re the best way to run a powerful model on your own desktop.
借助最新版本的 MLX，Muse Spark Open 已针对搭载 M5 Ultra 的 Mac Studio 进行优化，两者结合是在个人桌面上运行强大模型的最佳方式。
Apple’s unified memory architecture lets the M5 Ultra's integrated GPU use up to 512 GB per Mac Studio, so two machines can hold an entire model. And the chip’s 1.2 TB/s of memory bandwidth is fast enough to read the small slice each token needs. The result is two Mac Studios behave like one supercomputer with a terabyte of unified memory.
Apple 的统一内存架构允许 M5 Ultra 的集成 GPU 为每台 Mac Studio 使用高达 512 GB 的内存，因此两台机器即可容纳整个模型。而芯片 1.2 TB/s 的内存带宽足以快速读取每个 token 所需的小部分数据。结果是，两台 Mac Studio 表现得像一台拥有太字节统一内存的超级计算机。
We tested a range of agentic workflows running entirely on-device, using Muse Code powered by Muse Spark Open, including shipping real iOS changes to Ice Cubes (an open-source Mastodon client) — bug fixes, new features, full build — and driving DaVinci Resolve end to end, with the model reviewing footage frame by frame to find the strongest moments and crop them vertically.
我们测试了一系列完全在设备上运行的智能体工作流，使用由 Muse Spark Open 驱动的 Muse Code，包括向 Ice Cubes（一个开源 Mastodon 客户端）推送真实的 iOS 更改——修复错误、添加新功能、完整构建——以及端到端驱动 DaVinci Resolve，模型逐帧审查素材以找出最佳时刻并进行垂直裁剪。
Available Now
现已可用
We’re sharing this work because we want everyone to have access to the best models. The open-weight version of Muse Spark ships with custom, auto-tunable Helion GPU kernels in SGLang, delivering 1.03-3.95× kernel-level speedups on NVIDIA and AMD GPUs, and a 3% end-to-end speedup on AMD MI350.
我们分享这项工作，是因为我们希望每个人都能接触到最先进的模型。Muse Spark 的开源权重版本随附 SGLang 中自定义、自动可调谐的 Helion GPU 内核，在 NVIDIA 和 AMD GPU 上实现了 1.03-3.95 倍的内核级加速，并在 AMD MI350 上实现了 3% 的端到端加速。
Safety and Preparedness
安全与准备
Releasing open-weight models is part of how we bring the benefits of personal superintelligence to everyone, and part of how safety work advances across the field.
发布开源权重模型是我们将个人超级智能的好处带给每个人的方式之一，也是推动整个领域安全工作进展的方式之一。
Open weights also carry their own considerations, and Muse Spark Open has additional safety post-training for that reason. The results demonstrate strong safeguards across risk categories measured.
开源权重也带来其自身的考量，因此 Muse Spark Open 为此进行了额外的训练后安全处理。结果表明，在所评估的风险类别中，该模型具备强大的保障措施。
We evaluated the model in line with our Meta Superintelligence Scaling Framework, covering the most severe risks, and it met the thresholds our framework requires for release. Detailed results are in the Muse Spark Open model card, with a full evaluation report available soon.
我们按照 Meta 超级智能扩展框架对模型进行了评估，涵盖了最严重的风险，并且它达到了我们的框架所要求的发布阈值。详细结果见 Muse Spark Open 模型卡片，完整的评估报告即将发布。
License
许可证
We’re releasing the weights under the Muse Model License with limited commercial terms, which allow many developers to use the model for free.
我们在 Muse Model License 下发布权重，附带有限的商业条款，允许许多开发者免费使用该模型。
Get Started
入门指南
You can customize Muse Spark Open for your use case by leveraging PyTorch’s TorchTitan to further tune the model. To run it on Apple silicon, use MLX, which pulls the weights directly from Hugging Face.
您可以利用 PyTorch 的 TorchTitan 来定制 Muse Spark Open 以适应您的用例，从而进一步调整模型。要在 Apple Silicon 上运行，请使用 MLX，它直接从 Hugging Face 拉取权重。
Muse Spark Open is available now through Hugging Face: https://huggingface.co/collections/meta-models/muse-spark-open
Muse Spark Open 现已通过 Hugging Face 提供：https://huggingface.co/collections/meta-models/muse-spark-open
MLX Adapter for running Muse Spark Open on Mac Studio.
用于在 Mac Studio 上运行 Muse Spark Open 的 MLX 适配器。
Hosted Inference Partners: Together AI, Fireworks, Baseten, Modal, Groq, Nebius, and OpenRouter.
托管推理合作伙伴：Together AI、Fireworks、Baseten、Modal、Groq、Nebius 和 OpenRouter。
Built or published something with Muse Spark Open? Share with us on the Hugging Face Forums — we’d love to see what you create.
使用 Muse Spark Open 构建或发布了什么内容？请在 Hugging Face 论坛上与我们分享——我们很乐意看到您创造的作品。
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力