Technology
November 10, 2026 1017 words

Meta Muse Ember, Muse Spark Open and Spaces: Meta's Push Into Cheap, Agent-Ready AI

Hams AI Tech
Professional Content Team
Expert in AI Prompts & Professional Tools

Meta's Muse family has quickly become one of the most important lines of AI models in 2026, and the past several weeks brought three notable releases. In October, Meta published open weights for Muse Spark Open, a large multimodal model built for coding and agents. On November 6, it introduced Spaces in the Muse app, which builds custom planners, trackers and dashboards from a description. On November 9, it launched Muse Ember, its smallest and most efficient Muse model, priced aggressively for the repetitive work that AI agents do all day.

Muse Ember: Built for Agent Workloads

Most of what an AI agent does is not glamorous. It calls tools, extracts data, routes requests, handles small coding subtasks and parses long documents. Muse Ember is designed for exactly that work.

Key specifications:

    • Input: text, image, audio and video.
    • Context window: one million tokens.
    • Output: up to 512,000 tokens.
    • Pricing: 0.10 dollars per million input tokens, 0.50 dollars per million output tokens and 0.004 dollars per million cached input tokens.
    • Availability: muse-ember-1.0 on the Meta Model API and Muse Code, with Responses API support, and also on Google Cloud.

Meta says Ember is the first model it trained using Muse Wonder Full, its larger internal system, which helps explain how such a small model can perform well on practical tasks.

Why Ember's Price Matters

At ten cents per million input tokens, Ember sits among the cheapest capable multimodal models available, alongside Tencent's Hy4 flash. That opens up use cases that would be too expensive with premium models:

    • Processing every document in a large archive rather than a sample.
    • Running an agent continuously throughout the day to monitor inboxes or systems.
    • Extracting structured data from audio and video recordings at scale.
    • Using a cheap first pass to route only complex work to more expensive models.

Professionals building such pipelines can find practical templates in our AI automation resources and n8n workflow collection.

Muse Spark Open: A Large Open-Weight Model

In October, Meta released open weights for Muse Spark Open, a natively multimodal mixture-of-experts model with 816 billion total parameters, 42 billion active per token and a one million token context window. It is built for coding and agentic work. Weights are on Hugging Face, and hosted APIs are available through OpenRouter, Groq, Together AI and others.

For organizations that prefer to control their AI stack, Spark Open joins a growing list of large open models, covered in our open-weight models roundup.

Spaces in Muse: Apps From a Sentence

Spaces may be the most interesting release for everyday professionals. In the Muse app, you describe an idea and Muse builds a custom planner, tracker, dashboard, game or project hub. You then keep refining it in chat. Spaces can sit in the sidebar or main tabs, can be scheduled or hidden and can update in the background from connected apps. It is available everywhere the Muse app is.

This puts Meta in direct competition with workspace tools. Many professionals already use structured templates for the same purpose. If you prefer a proven, ready-made system, compare Spaces with our Notion templates, such as the Legal Project Management template or the freelance business tracker. Ready-made templates often encode best practices that a quick prompt might miss.

Muse Agents and the Web

Meta's Muse agent has also embraced open standards. According to Sierra, Meta's Muse agent checks for a Poppy file first on every site it visits, part of the new Personal Agent Protocol for how personal agents interact with businesses. For companies, that is a signal to prepare websites for agent visitors, not only human ones.

Practical Uses for Professionals

Lawyers and Paralegals

Ember's long context and low price make it suitable for first-pass document review and timeline extraction. Pair it with workflows such as the paralegal task automation suite and review anything that matters.

Accountants

Use Ember to extract figures from statements, receipts and audio notes, then send exceptions to a stronger model. See accounting templates and internal audit resources.

Marketers and Creators

Ember accepts video and audio, making it useful for summarizing recordings, generating captions and repurposing content. Explore content creation resources and social media growth tools.

Getting Started With Meta AI

If you are new to Meta's AI ecosystem, the Meta Muse Complete A to Z Guide walks through the Muse app, models and practical workflows. You can also browse our Meta AI section and the Meta AI collection.

Limitations

Small models trade depth for speed and price. Ember is not intended for complex legal reasoning or high-stakes financial analysis. Use it for structured, repeatable tasks with clear instructions, and escalate difficult items to a stronger model. As always, check where data is processed before sending confidential material. Spaces, meanwhile, are best for personal and team organization; for client records or regulated data, use systems your organization has approved and can audit.

Frequently Asked Questions

How much does Muse Ember cost?

0.10 dollars per million input tokens and 0.50 dollars per million output tokens, with cached input at 0.004 dollars per million.

Is Muse Ember open weight?

Ember is offered through the Meta Model API, Muse Code and Google Cloud. The open-weight model in the family is Muse Spark Open.

What can Spaces build?

Planners, trackers, dashboards, games and project hubs that you refine in chat and that can update from connected apps.

Can Ember handle video?

Yes. It accepts text, image, audio and video input with a one million token context.

The Bottom Line

Meta is competing on two fronts: very cheap models for agent workloads and consumer-friendly tools that build apps from a description. For professionals, Ember lowers the cost of automation and Spaces lowers the barrier to building simple tools. Combine them with proven templates and prompts for the best results, and measure the time saved on each workflow so you know where the new tools genuinely help.

Related AI News

Topics & Keywords
Meta Muse Ember Muse Spark Open Meta AI Muse app Spaces in Muse open weights