Meta 发布开源权重模型 Muse Spark Open，对标 Qwen、GLM 与 Kimi
原标题：Muse Spark Open is Meta's answer to Qwen, GLM, and Kimi in the open-weight race
AI 摘要
Meta 发布开源权重模型 Muse Spark Open，称其为美国最强开源权重模型，面向编程与智能体任务，拥有 8160 亿参数（激活 420 亿）、最高一百万 token 上下文，并支持文本、图像、视频和音频。Meta 表示其在 JobBench、MCP-Atlas、HLE-Diamond、ZeroBench 等基准上多个类别领先 Qwen 3.8、GLM 5.3 和 Kimi K3，但仍明显落后于其付费的 Muse Wonder Full 模型。该模型针对苹果 M5 Ultra 优化，可在两台各 512 GB 统一内存的 Mac Studio 上本地运行，权重已发布于 Hugging Face，许可免费但需标注“Built with Muse Spark Open”，年收入超 2000 万美元且想以 API 形式提供该模型的服务商需与 Meta 另行签约。
正文节选
Muse Spark Open is Meta's answer to Qwen, GLM, and Kimi in the open-weight race After its Llama series, Meta is returning to open-weight territory with Muse Spark Open. Meta calls it the most capable US open-weight model, built for coding and agentic tasks, with 816 billion parameters (42 billion active), up to one million tokens of context, and support for text, image, video, and audio. On benchmarks like JobBench, MCP-Atlas, HLE-Diamond, and ZeroBench, Meta says it scores well ahead of Qwen 3.