Muse Spark Open vs DeepSeek V4.1 Flash

Pyyan / Compare / Muse Spark Open vs DeepSeek V4.1 Flash

2 of 5

Open-Weight Models · verified 20 Oct 2026

Muse Spark OpenMetacurrent

DeepSeek V4.1 FlashDeepSeekcurrent

Add another3 slots left

Specification	Muse Spark Open	DeepSeek V4.1 Flash
Summary	The largest American open weights, under a licence with a revenue trigger.	Twice the parameters of the Flash it replaces, and cheaper to run.
Context	1M	1M
Input ($/Mtok)	Weights only	$0.3
Output ($/Mtok)	Weights only	$1.2
Licence	Muse Model License	MIT
Max output	Not published	256K
Parameters	816B total, 42B active	552B total, 8B / 16B active
Released	15 Oct 2026	10 Sep 2026, refreshed 17 Oct
Model ID	meta-models/Muse-Spark-Open-816B-A42B	deepseek-flash
Category	Open-Weight Models	Open-Weight Models
Official	Meta ↗	DeepSeek ↗

Highlighted rows are where these differ.

Text, image, video and audio in, text out, with a 1M token context
Muse Model License, not open source: attribution required, and a separate agreement above $20m of model-service revenue
About 1.5TB in BF16, or two Mac Studios at 4-bit

Best for multimodal work you host yourself.

552B total, 8B active while reading the prompt and 16B while generating, with a 1M token context
DeepSeek says it beats V4-Pro; V4-Pro requests route to it from 14 September 2026
MIT licence, native image input, and a KV cache about a quarter the size of V4-Flash's
Refreshed on 17 October as V4.1-Flash-1017, still MIT, improving on all eleven benchmarks DeepSeek published

Best for cheap agentic coding with images.