Muse Spark Open model now available on Amazon SageMaker JumpStart
Amazon SageMaker JumpStart now offers Meta's Muse Spark Open, an open-weight model optimized for agentic coding workflows, supporting long-running and multimodal engineering tasks.
Meta's Muse Spark Open is now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models for AWS customers. Optimized for agent-based coding workflows, Muse Spark Open is an open-weight model designed for long-running coding agents and multimodal engineering workflows. It uses a Mixture-of-Experts architecture with 816B total parameters and 42B active parameters per token, maintaining large-model capabilities while reducing compute per token. With a 1M token context window, it can process entire projects (including dependencies, legacy code, and thousands of files) at once. Accepting text, images, documents, audio, and audiovisual inputs, it outputs text. SageMaker JumpStart offers two variants: BF16 (Muse-Spark-Open-816B-A42B) and NVFP4 quantized (Muse-Spark-Open-816B-A42B-NVFP4), where the NVFP4 variant reduces memory usage by about 1/4 for cost-effective deployment. Using SageMaker JumpStart, customers can deploy Muse Spark Open in a few clicks, with the endpoint running in their AWS account for full control over source code and prompts. To use this model, access the SageMaker JumpStart model catalog in the SageMaker console or deploy the model to an AWS account using the SageMaker Python SDK. For details on deploying and using foundation models with SageMaker JumpStart, see the Amazon SageMaker JumpStart documentation.
Why it matters
Available for developers and software engineering teams to implement agentic coding workflows on AWS infrastructure.