Muse Wonder
Muse Wonder is Meta's frontier reasoning model, built for complex personal and professional agentic tasks. It reads text, images, video, audio and PDF documents and returns text, and with a 1M-token context window it can hold an entire mixed-media folder in one conversation. Now on Hoodgen AI.
Technical Specifications
Live data fetched from the OpenRouter model registry — always up to date.
What is Muse Wonder?
Muse Wonder is Meta's entry at the frontier tier, aimed squarely at agentic work that mixes formats. Where most text models stop at text and images, Muse Wonder also takes video, audio and PDF input, so a single prompt can reference a meeting recording, a slide deck and a spreadsheet and ask for one grounded answer. It carries a 1,048,576-token context window and up to 943,718 tokens of output, which is unusual even among frontier models.
Features & Capabilities
Five Input Types
Text, images, video, audio and PDF documents in one model.
1M Context
A whole folder of mixed media stays in view for the session.
Very Long Output
Up to 943,718 output tokens - among the largest ceilings available.
Frontier Reasoning
Agentic planning depth to match the flagship tier from Meta.
Personal and Pro Agents
Built for assistants that work across a person whole workload.
Meta at the Frontier
Meta back in the top bracket of frontier model builders.
Who Is Muse Wonder For?
Mixed-Media Analysis
A recording, a deck and a report cross-referenced in one pass.
Research Assistants
Agents that read widely and synthesise across formats.
Long-Form Generation
Outputs measured in hundreds of thousands of tokens.
Personal Workflows
Assistants that hold context across many unrelated tasks.
Muse Wonder Pricing
- $4.00 per million input tokens, $20.00 per million output tokens
- Unusually large output ceiling among frontier models
- Run it from your Hoodgen free credits, no credit card required
- Higher allowances on Hoodgen AI paid plans
What Makes Muse Wonder Different?
Muse Wonder is the model to reach for when the input is not plain text. Video, audio and PDF sit alongside text and images in the same context, which means a question like how does what was said in this meeting match the numbers in this deck can be asked directly rather than assembled by hand first.
The other headline is the output ceiling. Frontier models commonly cap output around 128,000 tokens; Muse Wonder goes to 943,718. That matters for work that produces rather than answers: long reports, extended transcripts, large generated documents, or a summary of a corpus that is itself enormous.
At $4 input and $20 output per million tokens it is priced with the other frontier flagships, and the patterns that keep cost sane are the same: do the exploratory passes on a cheaper model and bring Muse Wonder in for the analysis that needs every format at once.
Muse Wonder Performance & Pricing
Muse Wonder sits at the top of Meta model lineup and is one of the few models that accepts video and audio as well as text, images and PDFs.
What Can You Do With Muse Wonder?
Point Muse Wonder at anything that arrives in more than one format. It is the least awkward way to reason over a meeting recording, a set of scanned documents and a spreadsheet without preprocessing each into text yourself.
Best Use Cases
- Cross-checking a recording against a written report
- Reading scanned PDFs and answering questions about them
- Producing very long deliverables in one pass
- Summarising a mixed-media research folder
- Agents that must handle whatever format arrives
Muse Wonder Privacy & Data Policy
Muse Wonder is developed by Meta and served through OpenRouter. Prompts and completions are processed by Meta model providers under their terms.
Because this model accepts audio and video, it is worth being deliberate about what you upload. Avoid sending confidential recordings or personal documents you would not share with an external provider.
Muse Wonder vs Other Models
Frequently Asked Questions
Reason across every format
Chat with Muse Wonder on Hoodgen - sign in and start in seconds.
Chat with Muse Wonder