Meta Platforms released Muse Spark Open, an open-weight artificial intelligence model aimed at developers building coding tools and AI agents. The company positions it as the most capable open-weight model developed in the United States so far.
Unlike a fully closed model, Muse Spark Open lets developers download its trained parameters, or ‘weights,’ and run or adapt them on their own systems. The release expands developer choice while Meta retains control over its most advanced models.
Muse Spark Open is a coding-optimized model designed for complex software engineering, multi-step reasoning and long-running agentic tasks. Meta says it can handle work that requires planning, tool use and sustained execution rather than single-turn responses.
The model is natively multimodal. It accepts text, images, audio, and video as input and generates text output, making it suitable for customer-service agents, developer assistants, and enterprise automation. Muse Spark Open features 816 billion total parameters but activates about 42 billion parameters for each token. The model also supports context window of one million tokens. Thus, allowing it to process lengthy codebases, large documents, and extended task histories in a single session.
Meta collaborated with Apple to optimize Muse Spark Open for Mac hardware. Developers can run the model locally on two Apple M5 Ultra Mac Studio computers using Apple’s MLX framework. Meta made the model available through API providers including
Baseten
Fireworks AI
Groq
Modal
Nebius
OpenRouter
Together AI
Developers can download the weights from Hugging Face, Amazon SageMaker AI, Google Cloud Model Garden and Microsoft Foundry Models. Meta says, “It built safeguards into the release, but the license still restricts how developers may deploy it.” The arrangement is notable because it brings a very large AI model to desktop hardware. The model is free to download under the Muse Model License, which permits limited commercial use.
Also Read: Meta’s Muse Wonder AI Challenges ChatGPT and Claude: What Makes It Different?