Images & video
Muse Wonder image & video capabilities: understanding vs generation
See what Muse Wonder can do with images and videos, how text output differs from media generation, and prompts for your own provider tests.
Last verified · Independent editorial guide
Wonder accepts image and video inputs and returns text. Meta documents Muse Image as a separate image-generation model. This page links real official examples and supplies test prompts; it does not fabricate model outputs or host live inference.
Image understanding is not image generation
Official documentation· Last verified
An image-understanding request can ask a model to describe a chart, inspect a screenshot or reason about a photograph. The result is text about that input, not a newly generated image.
For visual tasks, supply an image you have the right to use and ask for evidence tied to visible details. Verify small text, numerical values and spatial claims manually; a confident description is not proof of correctness.
Video input needs a question and a checking method
Official documentation· Last verified
A useful video test asks for a specific event, sequence or discrepancy, with supporting time references when the endpoint supports that workflow. Check the exact media formats, size limits and request syntax in the provider documentation.
The prompts below are reproducible instructions for your own tests, not transcripts from Wonder. No image is uploaded and no paid model call is made on this page. Use the official provider to run the test.
Watch the official examples with their limits in mind
Official documentation· Last verified
Meta’s release page includes demonstrations of coding and agents working with visual information. They illustrate the provider’s selected workflows rather than a controlled independent comparison.
If you need image creation or editing, inspect Muse Image’s own documentation. The Wonder name, a video input option or an impressive agent demo does not establish a video-generation capability.
Test with your own material
These are test prompts, not model outputs. Run them with your material at the provider; this site does not receive your files.
Image understanding test
Inspect this interface screenshot. List the visible navigation, primary actions and error messages. Tie each finding to visible evidence. Mark unreadable text as uncertain rather than inventing it.
Video understanding test
Summarize the sequence of actions in this video. Support each step with a visible event and a time reference where supported. Distinguish observations from inference and do not invent missing details.
Frequently asked questions
Editorial synthesis · Cited sources· Last verified
Can Muse Wonder generate images?
The checked model documentation lists text output for Wonder. Muse Image is Meta’s separate image-generation and editing model.
Can I upload a video on this website?
No. This independent guide links provider documentation and examples. It does not accept media or run paid model inference.
Are the sample prompts real model results?
No. They are original test instructions you can run with your own provider account. No generated answer is presented as a measured result.
Sources and verification
Provider pricing, model capabilities and access are distinguished from independent evaluations. The date records an editorial check; data is not refreshed automatically.