Muse Spark 1.2

Closed Source

Closed Weights

Parameters

Context Length

1.05M

Modality

Multimodal

Architecture

Dense

License

Release Date

Knowledge Cutoff

Evaluation Benchmarks

Rank

#14

Benchmark	Score	Rank
General Text high Text Arena	1499	⭐ 5
General Text Standard Text Arena	1499	⭐ 5
Reasoning high LiveBench Reasoning	0.90	⭐ 7
Reasoning Standard LiveBench Reasoning	0.90	⭐ 7
General high LiveBench Global	0.78	14
Agentic Coding high LiveBench Agentic	0.58	14
LiveBench Average high LiveBench Average	0.78	16
LiveBench Average Standard LiveBench Average	0.78	16
Data Analysis high LiveBench Data Analysis	0.76	26
Mathematics high LiveBench Mathematics	0.91	27
Mathematics Standard LiveBench Mathematics	0.91	27
Web Development WebDev Arena	1534	28
Coding LiveBench Coding	0.78	31
Agent Arena high Agent Arena	0.85	46

Rankings

Overall Rank

#14

Coding Rank

#12

About Muse Spark 1.2

Muse Spark 1.2 is a frontier multimodal reasoning model developed by Meta for complex agentic workflows and multi-step problem solving. It accepts text, images, video, audio, and documents across a 1M token context window.

Technical Specifications

Detailed architecture specifications are limited for proprietary models.

About Llama 4

Meta's Llama 4 model family implements a Mixture-of-Experts (MoE) architecture for efficient scaling. It features native multimodality through early fusion of text, images, and video. This iteration also supports significantly extended context lengths, with models capable of processing up to 10 million tokens.