None
None
Long Story Video Skill
Turn your idea into 10s–10min video—AI writes scripts, prompts, and generates footage automatically.
Ads Video Skill
Generate professional ads and sales videos—AI auto-generates scripts, prompts, and footage.
3D Science Explainer Video Skill
Convert scientific concepts into stunning 3D explain animations
Feedback
freeTrialImage.bannerPity
freeTrialImage.upgradeUnlock
- ✓freeTrialImage.benefitHd
- ✓freeTrialImage.benefitWatermark
- ✓freeTrialImage.benefitUnlimited
Muse Spark Open
Muse Spark Open brings open-weight multimodal AI to your desk: 1M-token context, local inference, and agentic video editing on Apple silicon.
All Tools
Discover our comprehensive AI-powered animation toolkit
MiniMax H3
MiniMax H3 AI Video Generator
Seedance 2.5
The Future of AI Video Is Here.
Seedance 2.0
The Future of AI Video Is Here.
Veo3.1
Create Stunning Videos with Veo3.1
FLUX 3 Video Generator
Kling 3.0
Next-Gen AI Video Generator
Grok Video Generator
Create Videos from Text or Images with AI
MiniMax H3 video generator
MiniMax H3 AI Video Generator
Own Your AI Stack with Muse Spark Open
Muse Spark Open is released as open weights: a natively multimodal engine for coding, autonomous workflows and long-running tasks that you can host on your own hardware.
- Openly Licensed, US-DevelopedThis release stands as the most capable open-weight model from a US lab — 816 billion total parameters with 42 billion active, a context window spanning one million tokens, and multimodal input handled natively.
- Supercomputer-Class AI on a WorkstationFrontier models do not have to live only in data centers; Muse Spark Open can start on your desk when paired with two Apple M5 Ultra Mac Studios and the latest MLX runtime.
- Agentic Workflows Tested LocallyMuse Code running on Muse Spark Open delivered real iOS fixes and features to Ice Cubes, then controlled DaVinci Resolve from start to finish, including vertical crops of the best frames.
Deploy Muse Spark Open in Three Steps
Choose your path: local Apple silicon, a self-hosted inference engine, or a managed API — each route gets Muse Spark Open working for you.
What Muse Spark Open Can Do
A million-token context, native multimodality, desktop-class memory bandwidth and open weights — these are the traits that let Muse Spark Open handle agentic video work.
One-Million-Token Context
Muse Spark Open keeps a 1M-token window, so extended agentic sessions and multi-step reasoning chains retain earlier context.
Agentic Coding Across Long Horizons
Built for software engineering and autonomous loops, Muse Spark Open manages long-running jobs and end-to-end control of tools like DaVinci Resolve.
Multimodal by Design
Muse Spark Open ingests mixed media directly instead of relying on bolt-on adapters, enabling frame-by-frame review of footage.
Two Mac Studios as One Supercomputer
Unified memory allows each M5 Ultra GPU to address up to 512 GB, so two Mac Studios can hold the entire Muse Spark Open model as a single machine with a terabyte of pooled memory.
1.2 TB/s of Memory Bandwidth
The M5 Ultra delivers 1.2 TB/s of bandwidth — enough to feed the small per-token slice and keep Muse Spark Open responsive on a desktop.
Safety Post-Training Included
Open weights receive additional safety post-training: Muse Spark Open was evaluated against the Meta Superintelligence Scaling Framework and passed its release threshold across measured risk categories.
Muse Spark Open Questions, Answered
Answers to common questions about running, licensing and downloading Muse Spark Open.
Which setup can run Muse Spark Open on my own machine?
Two Apple M5 Ultra Mac Studios with the latest MLX build — each supports up to 512 GB of unified memory, and together they provide roughly a terabyte for the full model.
What are the model size and context limits?
Muse Spark Open has 816B total parameters with 42B active, a 1M-token context window and native multimodality, designed for coding, agentic work and long-horizon tasks.
Where can I get the weights or call the model through an API?
You can get the weights through Amazon SageMaker AI, Google Cloud Model Garden, Hugging Face and Microsoft Foundry Models, while hosted APIs are offered by Baseten, Fireworks AI, Groq, Modal, Nebius, OpenRouter and Together AI.
Which inference engines and hardware are supported?
Choose from MLX, llama.cpp, Unsloth, SGLang, vLLM, TensorRT-LLM or Transformers — tuned for NVIDIA, AMD, Intel and Apple hardware, with PyTorch underneath when you need deeper control.
What license covers the released weights?
Licensing follows the Muse Model License; its limited commercial terms still leave ample room for a broad range of developers to run Muse Spark Open at no cost.
How was Muse Spark Open evaluated for safety before release?
The model was assessed with the Meta Superintelligence Scaling Framework across the most severe risk categories, given additional safety post-training for open weights, and met the release thresholds — full details live in the model card.
Start Building with Muse Spark Open Today
Grab Muse Spark Open from Hugging Face, query it through hosted inference providers, and show the community what you create in the Hugging Face Forums.