Muse Spark Open

Name: Muse Spark Open
Author: Meta

Multimodal

Meta

Muse Spark Open is Meta Superintelligence Labs' open-weight natively multimodal mixture-of-experts model, released 2026-10-15 with 816B total and 42B active parameters. It reads text, image, video and audio with a 1,048,576-token context window and configurable reasoning effort, and is positioned for coding and agentic work. Weights ship on Hugging Face under the Muse Model License with a $20M revenue threshold for model-as-a-service use; Together AI hosts it at $1.25/$4.25 per million tokens.

Key Specifications

Parameters

816.0B

Context

1.0M

Release Date

October 15, 2026

Average Score

62.3%

API Documentation Research Paper Model Weights Results Blog

Timeline

Key dates in the model's history

Announcement

October 15, 2026

Last Update

October 20, 2026

Today

October 27, 2026

Technical Specifications

Parameters

816.0B

Training Tokens

Knowledge Cutoff

Family

Capabilities

MultimodalZeroEval

Pricing & Availability

Input (per 1M tokens)

$1.25

Output (per 1M tokens)

$4.25

Max Input Tokens

1.0M

Max Output Tokens

1.0M

Supported Features

Function CallingStructured OutputCode ExecutionWeb SearchBatch InferenceFine-tuning

Benchmark Results

Model performance metrics across various tests and benchmarks

Reasoning

Logical reasoning and analysis

GPQA

HF Muse Spark Open model card; xhigh. GPQA Diamond 89.0; catalog stores Diamond under gpqa. • Self-reported

89.0%

Other Tests

Specialized benchmarks

AA-Omniscience Index

HF Muse Spark Open model card; xhigh. AA Omniscience signed index 25 (−100..100); normalized_score left null. • Self-reported

12.5%

Artificial Analysis

HF Muse Spark Open model card table; xhigh reasoning. Artificial Analysis Intelligence Index reported as 39 on a 0-100 scale, stored as a fraction. • Self-reported

39.0%

BabyVision

HF Muse Spark Open model card; xhigh. BabyVision (w/ tools) 80.2. • Self-reported

80.2%

DeepSWE 1.1

HF Muse Spark Open model card; xhigh. DeepSWE 1.1 Preferred harness 60. • Self-reported

60.0%

GDPval-AA

HF Muse Spark Open model card; xhigh. GDPVal-AA v2 Elo 1401 (max_score 3000). • Self-reported

46.7%

Humanity's Last Exam (with tools)

HF Muse Spark Open model card; xhigh. HLE (w/ tools) 61.0. • Self-reported

61.0%

Job Bench

HF Muse Spark Open model card; xhigh. JobBench 61.8. • Self-reported

61.8%

MCP Atlas

HF Muse Spark Open model card; xhigh. MCP-Atlas 89.2. • Self-reported

89.2%

OSWorld-Verified

HF Muse Spark Open model card; xhigh. OSWorld Verified 80.4. • Self-reported

80.4%

SciCode

HF Muse Spark Open model card; xhigh. SciCode 58.5. • Self-reported

58.5%

SimpleVQA

HF Muse Spark Open model card; xhigh. SimpleVQA (w/ tools) 73.6. • Self-reported

73.6%

SWE Atlas - Codebase QnA

HF Muse Spark Open model card; xhigh. SWE-Atlas-QnA Preferred harness 58; mapped to swe-atlas-codebase-qna (same mapping as muse-spark-1.3). • Self-reported

58.0%

Terminal-Bench 2.1

HF Muse Spark Open model card; xhigh. Terminal-Bench 2.1 Terminus 2 agent 76.4. • Self-reported

76.4%

Toolathlon-Verified

HF Muse Spark Open model card; xhigh. Toolathlon Verified 72.8. • Self-reported

72.8%

WorldVQA

HF Muse Spark Open model card; xhigh. WorldVQA (w/ tools) 49.4. • Self-reported

49.4%

ZEROBench

HF Muse Spark Open model card; xhigh. ZeroBench (w/ tools) 50.0. • Self-reported

50.0%

License & Metadata

License

muse_model_license

Announcement Date

October 15, 2026

Last Updated

October 20, 2026

Compare Muse Spark Open

All comparisons

vs Muse Glimmer-30B vs Muse Wonder Full vs Muse Spark 1.2 vs Muse Spark 1.3 vs IBM Granite 4.2 30B vs GPT-6.1 Sol vs North Micro Vision Instruct vs Nemotron 3.5 Super VL (120B A12B)

Similar Models

All Models

Llama 4 Scout

Meta

MM109.0B

Best score:0.8 (MMLU)

Released:Apr 2025

Price:$0.18/1M tokens

Llama 4 Maverick

Meta

MM400.0B

Best score:0.9 (MMLU)

Released:Apr 2025

Price:$0.27/1M tokens

Muse Spark

Meta

Best score:0.9 (TAU)

Released:Apr 2026

Muse Glimmer-30B

Meta

MM29.6B

Best score:0.8 (GPQA)

Released:Aug 2026

Llama 3.1 405B Instruct

Meta

405.0B

Best score:1.0 (ARC)

Released:Jul 2024

Price:$3.50/1M tokens

Command A+

Cohere

MM218.0B

Best score:0.8 (TAU)

Released:May 2026

Step-3.5-Flash

StepFun

MM196.0B

Best score:0.9 (TAU)

Released:Feb 2026

Price:$0.10/1M tokens

Inkling-Small

Thinking Machines Lab

MM276.0B

Best score:0.9 (GPQA)

Released:Jul 2026

Price:$0.30/1M tokens

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.