Meta/Mixture of Experts

Muse Spark Open

Name: Muse Spark Open
Author: Meta

Muse Spark Open — Meta Superintelligence Labs' open-weight 816B MoE with 42B active and a 1M context, natively multimodal across images, audio and video, built for coding and agentic work. Ships with extra safety post-training relative to the hosted Muse Spark.

chatcodingreasoningmultilingualvisionaudioagentictool_use

822.04B

Parameters (42B active)

1024K

Context length

Benchmarks

Quantizations

10K

HF downloads

Architecture

MoE

Released

2026-10-15

Layers

KV Heads

Head Dim

Family

muse

Quantization Options

Context length:

Quant	Bits	VRAM @ 16K	Quality
IQ2_XXS	2.38	246.1 GB 245.0 + 1.1 KV	low
IQ2_M	2.93	302.6 GB 301.6 + 1.1 KV	low
Q2_K	3.16	326.3 GB 325.2 + 1.1 KV	low
IQ3_XXS	3.25	335.5 GB 334.4 + 1.1 KV	low
IQ3_XS	3.5	361.2 GB 360.1 + 1.1 KV	low
Q3_K_S	3.64	375.6 GB 374.5 + 1.1 KV	low
IQ3_M	3.76	387.9 GB 386.8 + 1.1 KV	low
Q3_K_M	4	412.6 GB 411.5 + 1.1 KV	low
Q3_K_L	4.3	443.4 GB 442.3 + 1.1 KV	moderate
IQ4_XS	4.46	459.8 GB 458.8 + 1.1 KV	moderate
Q4_K_S	4.67	481.4 GB 480.4 + 1.1 KV	moderate
Q4_K_M	4.89	504.0 GB 503.0 + 1.1 KV	good
Q5_K_S	5.57	573.9 GB 572.8 + 1.1 KV	good
Q5_K_M	5.7	587.3 GB 586.2 + 1.1 KV	good
Q6_K	6.56	675.6 GB 674.6 + 1.1 KV	excellent
Q8_0	8.5	875.0 GB 873.9 + 1.1 KV	lossless
FP16	16	1645.6 GB 1644.6 + 1.1 KV	lossless

Select your GPU above to see speed estimates and compatibility for each quantization.

Too big for a single GPU — plan a multi-GPU deployment

Even the lightest quant needs ~246 GB. Size GPUs, replicas, TCO and scaling for a production setup. Open in Enterprise →

▸ READY TO RUN THIS?RENT BY THE HOUR

RENT A GPU AND RUN MUSE SPARK OPEN NOW

Rent on Vast.ai →Or RunPod →

Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.

Community Ratings

Loading ratings...

Benchmarks (6)

GPQA Diamond89.0

AA Long Context78.3

Terminal-Bench 2.176.4

SciCode58.4

HLE43.9

Terminal-Bench 48.1

Run this model

▸Easiest way to get started·Beginners

DOCS ↗

curl -fsSL https://ollama.com/install.sh | sh

$ollama run muse:822b-q4_K_M

Tag may need adjustment — check ollama.com/library/muse for available tags.

▸ SETUP GUIDE

Auto-setup with fitmyllm CLI

Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.

pip install fitmyllmthen run fitmyllmLearn more

Auto-detect GPULive tok/s in chatSpeed benchmarks9 inference engines

HuggingFace GGUF Downloads Build Hardware

Find the best GPU for Muse Spark Open

Build Hardware for Muse Spark Open

QUANT	BPW	VRAM	QUALITY
IQ2_XXS	2.38	245.0 GB	65%
IQ2_M	2.93	301.6 GB	75%
Q2_K	3.16	325.2 GB	78%
IQ3_XXS	3.25	334.4 GB	82%
IQ3_XS	3.5	360.1 GB	84%
Q3_K_S	3.64	374.5 GB	85%
IQ3_M	3.76	386.8 GB	86%
Q3_K_M	4	411.5 GB	88%
Q3_K_L	4.3	442.3 GB	90%
IQ4_XS	4.46	458.8 GB	92%
Q4_K_S	4.67	480.4 GB	93%
Q4_K_M	4.89	503.0 GB	94%
Q5_K_S	5.57	572.8 GB	96%
Q5_K_M	5.7	586.2 GB	96%
Q6_K	6.56	674.6 GB	97%
Q8_0	8.5	873.9 GB	100%
FP16	16	1644.6 GB	100%