Muse Spark Open, the Meta model in numbers

Overview

Muse Spark Open is a language model from Meta, released on October 15, 2026. It is a very recent release. Its weights are published: you can download it and run it on your own servers, or use one of the 15 API providers that list it on models.dev. It has a reasoning mode: it can "think" before answering, which helps with multi-step problems but uses more tokens, and therefore more money.

We file it under open-weight models (68 models in this category on the site).

How much does Muse Spark Open cost?

Muse Spark Open costs $1.25 per million input tokens and $4.25 per million output tokens (median of 15 providers, checked on October 22, 2026).

To put it in perspective: reading 20,000 tokens (about thirty pages) and writing 2,000 costs roughly $0.034; a thousand of those tasks, $33.5.

That is 4.3 times the median output price of open-weight models ($1).

To estimate your own budget, use the agent cost calculator.

What do independent benchmarks say?

According to results published by Epoch AI (checked on October 22, 2026), Muse Spark Open has a capability index of 75.2 out of 100, computed on 4 benchmarks from our basket.

It is not among the top 30 of the weekly ranking.

• GPQA Diamond (PhD-level science questions): 88.6% — the best tracked model, Gemini 4 Argon, reaches 96.4%
• OTIS Mock AIME (competition mathematics problems): 98.1% — the best tracked model, GPT-5.5 Pro, reaches 100%
• FrontierMath (very hard research-level mathematics): 69.8% — the best tracked model, GPT-6 Astra, reaches 93.7%
• APEX-Agents (professional tasks carried out by an agent): 33.4% — the best tracked model, Gemini 4 Argon, reaches 82.4%

Context and capabilities

Its context window reaches 1,048,576 tokens, roughly 1,573 pages of text at 500 words per page: enough to load a whole book or a mid-sized codebase. A single answer can run to 1,048,576 tokens.

It accepts as input: text, image and video. Its knowledge stops at January 4, 2026: beyond that, it needs web search or your documents.

It can call tools (functions, APIs) and return structured JSON output, which makes it a possible foundation for an AI agent.

The weights are published on Hugging Face (meta-models/Muse-Spark-Open-816B-A42B). See also our list of open-source models.

When to choose it

Given its category, Muse Spark Open is mainly aimed at projects where you want control over your data or hosting, or want to adapt the model to your field.

Its very large context window makes it worth considering for analysing large sets of documents in a single request.

Before adopting it, test it on your own examples: benchmarks measure standard tasks, not your job.

What could replace it?

In the same category, the models closest to Muse Spark Open are:

• GLM-5 (Zhipu AI (Z.ai), index 77.2) — see the comparison
• Kimi K2.5 (Moonshot AI, index 78.1) — see the comparison
• DeepSeek V4 Pro 0813 (DeepSeek, index 72.2) — see the comparison
• Qwen3.6 27B (Alibaba Cloud (Qwen), index 72.2) — see the comparison