Overall Capabilities and Architecture Mixture of Experts Perception Encoder Audio Encoder Long Context Inference Support Transformers Audio Inference Llama.cpp SGLang vLLM MLX Deploying Muse Spark Open with Inference Endpoints Fine-tuning with TRL Demos 3D Asset Generator Browser Interactions with OpenEnv and Hugging Face Sandbox Call your Browser Wrapping Up Muse Spark Open is an open-weight model from Meta. The mixture of experts model brings text, images, audio, and video into a model with approximately 816 billion parameters, around 42 billion active per token, and a context window of up to one million tokens. It generates text, giving developers a common interface for asking questions about different kinds of input.