Muse Spark Open model now available on Amazon SageMaker JumpStart

Meta's Muse Spark Open on Amazon SageMaker JumpStart

Meta's Muse Spark Open, a coding-optimized open-weight model, is now available on Amazon SageMaker JumpStart. This model is purpose-built for agentic coding workflows, supporting multi-step reasoning and tool calling for enterprise software engineering on AWS infrastructure.

Designed for long-running coding agents and multimodal engineering workflows, Muse Spark Open is built on a Mixture-of-Experts (MoE) architecture with 816B total parameters and 42B active per token. It offers a 1M-token context window, enabling it to handle entire projects without splitting tasks. The model accepts various inputs such as text, images, documents, audio, and audio-video, generating text output. It is trained on real developer workflows, planning, acting, observing results, and self-correcting over long horizons.

SageMaker JumpStart provides two variants: BF16 (Muse-Spark-Open-816B-A42B) and NVFP4 quantized (Muse-Spark-Open-816B-A42B-NVFP4), with the NVFP4 variant requiring about a quarter of the memory for cost-efficient deployment.

Customers can deploy Muse Spark Open with just a few clicks, with the endpoint running in their own AWS account for full control over source code and prompts.

What to do

• Navigate to the SageMaker JumpStart model catalog in the SageMaker console.
• Use the SageMaker Python SDK to deploy the model to your AWS account.
• Refer to the Amazon SageMaker JumpStart documentation for more information on deploying and using foundation models.

Source: AWS release notes

If you need further guidance on AWS, our experts are available at AWS@westloop.io. You may also reach us by submitting the Contact Us form.