estimate, except measured file sizes

Can I run Meta Muse Spark Open locally?

816B total, 42B active. Parameter count is the published figure.

Licence:

Muse Model License source

Context:

1,048,576 tokens

Modality:

text + image + video + audio in, text out

Released:

15 Oct 2026 source

MoE, 256 experts, 8 active per token, 1.8B perception encoder plus audio encoder. The card states 816B total / 42B active; Hugging Face metadata lists 822B including the vision and audio encoders. Licence (Muse Model License): products must display "Built with Muse Spark Open". Model-as-a-Service offerings by licensees whose group revenue exceeds US$20M in any 12 months need a separate agreement with Meta. Internal use and powering your own non-MaaS product are exempt. Gated download (accept terms on Hugging Face). Model card: "optimized to run on two Mac Studios with M5 Ultra using MLX". NVFP4 is for Blackwell GPUs only and needs at least 700GB of VRAM per the card. Vision projector (mmproj) is not a fit row: Q8_0 2.18 GB and BF16 4.10 GB, so add +2.2–4.1 GB for image/audio input. HF blog: 2-bit GGUF (304 GB) ran at 18 tok/s and oQ3e MLX at ~21 tok/s generation on a Mac Studio M3 Ultra 512GB (vendor/HF-reported, single test, not ours). Ollama offers muse-spark-open:cloud only (no local tag on 17 Oct 2026, https://ollama.com/library/muse-spark-open/tags). Supported in mainline llama.cpp (ggml-org GGUFs), vLLM, SGLang, mlx-lm/oMLX. Speed is not estimated: the formula streams every weight and understates sparse MoE.

Parameter source · Architecture source · Model page

speed: not estimated for MoE yet

Memory per quant

KV is at 8,192 tokens. Overhead is 0.8 GB. Both are estimates. A measured file size is labelled. The last column is the cheapest machine that fits the total fully.

Cheapest rig that fits

Q4, full: Nothing on our list of 32 machines fits this quant.

Q4, slow, partly in system RAM: Nothing on our list of 32 machines fits this quant.

UD-IQ1_S (smallest file), full: Mac Studio M5 Ultra 256GB, €12,500 (estimate)

UD-IQ1_S (smallest file), slow, partly in system RAM: Nothing on our list of 32 machines fits this quant.

Prices moved: see Building a local LLM rig in the RAMageddon.

On a specific GPU

• Meta Muse Spark Open on RTX 3060 12GB
• Meta Muse Spark Open on RTX 4060 Ti 16GB
• Meta Muse Spark Open on RTX 3090
• Meta Muse Spark Open on RTX 4090
• Meta Muse Spark Open on RTX 5080
• Meta Muse Spark Open on RTX 5090
• Meta Muse Spark Open on RX 7900 XTX
• Meta Muse Spark Open on Dual RTX 3090
• Meta Muse Spark Open on Dual RTX 4090