Choose a model for your next creation

Understand image, video and music inputs and the settings that change between models.

Updated October 8, 2026

Choose the output first

Start in Image for a still composition, Video for a moving scene, or Music for a soundtrack idea. The model menu changes with the selected tab. A useful first question is whether you need a finished asset or a visual draft: an image can help establish composition before spending a video creation on movement. This is a workflow choice, not a claim that one model is universally better than another.

Reference support is model specific

Some models accept a reference image while others work from text. Additional inputs, such as an end frame, are only available where supported. Check the upload control and options for the selected model. Do not assume that an image-to-video workflow is available just because a model name appears in the video menu. Unsupported settings are not a way to unlock additional capability.

Resolution and duration have tradeoffs

Use the lowest available resolution that lets you evaluate the idea before refining it. A longer clip asks the model to maintain the subject and scene over more frames. For the current free release, available Seedance models are limited to 480p; Seedance 2.5 is hidden. Other models have their own allowed settings. The studio shows compatible choices and validates submitted requests on the server.

Make a useful comparison

Keep the prompt, intended framing and subject similar when comparing outputs. Look at the details that matter for your project: faces, text, product shape, camera movement or music structure. A visually appealing homepage sample is not a benchmark or a promise of repeatable output. Model availability can change, so the live model selector is the source of truth for what you can use today.

Know which service you are using