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 runs locally: open weights, 816B parameters and a 1M-token window for agentic video edits on your own Mac.
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
A Frontier Model You Actually Own
Muse Spark Open ships as downloadable weights — a natively multimodal model for coding, agentic loops and long-running tasks that lives on hardware you control.
- Weights Anyone Can DownloadThe full checkpoint is public: Muse Spark Open carries 816B total parameters with 42B active, a million-token window and multimodal input built in from day one.
- A Supercomputer on Your DeskNo datacenter lease required — a couple of Apple M5 Ultra Mac Studios running the newest MLX build is enough to boot the entire model beside your monitor.
- Agentic Edits, Already Battle-TestedMuse Code shipped real iOS patches and features to Ice Cubes, then drove DaVinci Resolve start to finish and reframed the strongest shots vertically.
Launch Muse Spark Open in Three Steps
Three paths into Muse Spark Open — Apple silicon on your desk, a serving engine you control, or a hosted endpoint that answers in seconds.
Core Strengths of Muse Spark Open
A million-token context, true multimodal input, workstation memory bandwidth and public weights — these are the traits of Muse Spark Open that make multi-step video work possible.
A Million-Token Context
Long sessions stay coherent: earlier steps and reasoning chains remain in view even after hours of multi-step work inside Muse Spark Open.
Agentic Coding That Runs Long
Built for software engineering loops, it sustains long-horizon jobs and drives tools such as DaVinci Resolve end to end without losing the thread.
Multimodal by Design
Images, frames and text arrive through one native pathway instead of bolt-on adapters, which is exactly why frame-by-frame footage review works.
Two Macs Acting as One Machine
Unified memory lets each M5 Ultra GPU reach 512 GB, so a pair of Mac Studios pools roughly a terabyte and holds the whole Muse Spark Open model at once.
1.2 TB/s of Memory Bandwidth
That much throughput feeds the small active slice on every single token, keeping Muse Spark Open responsive on a desktop instead of sluggish.
Extra Safety Post-Training
The open release went through additional safety tuning, was scored against the Meta Superintelligence Scaling Framework and cleared every measured risk threshold before launch.
Muse Spark Open: Frequently Asked Questions
Answers on hardware, licensing, downloads and API access for Muse Spark Open and its open-weight release.
Which hardware runs Muse Spark Open locally?
Two Apple M5 Ultra Mac Studios on the latest MLX build — each offers up to 512 GB of unified memory, giving roughly a terabyte for the full checkpoint.
How big is the model, and what context does it support?
Muse Spark Open holds 816B total parameters with 42B active, a 1M-token window and native multimodality, aimed at coding, agentic loops and long-horizon jobs.
Where do I get the weights, or call the model through an API?
Grab the files from Amazon SageMaker AI, Google Cloud Model Garden, Hugging Face or Microsoft Foundry Models; hosted endpoints run at Baseten, Fireworks AI, Groq, Modal, Nebius, OpenRouter and Together AI.
Which serving engines and chips work with it?
Choose MLX, llama.cpp, Unsloth, SGLang, vLLM, TensorRT-LLM or Transformers — tuned for NVIDIA, AMD, Intel and Apple silicon, with PyTorch underneath when you need deeper control.
What licence covers the released weights?
The Muse Model License governs use; its limited commercial terms still leave broad room for developers to run Muse Spark Open at no cost.
How was safety evaluated before the release?
It was scored against the Meta Superintelligence Scaling Framework across the most severe risk categories, received extra safety post-training for open distribution and cleared the release thresholds — the model card has the details.
Start Building with Muse Spark Open Today
Pull the Muse Spark Open checkpoint from Hugging Face, query it through partner endpoints, then post what you make in the community forums.