Sora alternatives by API: eight Sume video ids compared

If you are looking for a Sora alternative you can call from code, Sume's Video Router lists eight explicit video ids: seedance-2.5, seedance-2, wan-3.0, minimax-h3, minimax-h3-max, higgsfield-genjutsu, h3-max-recast and gemini-omni-flash-1.1. They are not interchangeable. The first thing to match is clip length, because the ids range from 3 seconds to 30 seconds, and the second is whether you need references or an edit mode.

This page makes no claims about what Sora does or does not offer, because OpenAI's own pages were not fetchable when it was written. It is purely a map of the Sume side, taken from the Video Router and Video generation docs. For a prompt-level migration, see the linked guide on rewriting a Sora prompt for Omni Flash.

The eight ids

The table lists the limits the docs state per id. Two docs lines apply to every row: Sume bills provider list price times 1.25 on every Video Router model, and any model not named with a longer limit is capped at 15 seconds.

Matching a job to an id

For a single clip over 15 seconds, only seedance-2.5, wan-3.0, higgsfield-genjutsu and h3-max-recast apply, and the last two need a source video. For a 10-second clip with sound at high resolution, gemini-omni-flash-1.1 reaches 4K and always includes native audio. For an existing clip you want to change rather than regenerate, Omni Flash has a video_url edit mode, and h3-max-recast swaps the people in a source video using one photo per person.

If you do not want to choose, POST /v1/videos with model: "sume/auto" lets Sume pick. The docs say Auto defaults to 720p and 8 seconds, with 3 to 10 second clips at 16:9 or 9:16, and responses echo sume/auto without disclosing which family ran. Use a pinned id when you need repeatable output or a specific length.

Reference limits differ too

If your Sora workflow fed in images or clips, check reference support before choosing. Only models whose supported_input_references lists a type accept that type. Audio and video references are honored by the Seedance 2.x models, Wan 3.0, MiniMax H3 and MiniMax H3 Max. Gemini Omni Flash 1.1, higgsfield-genjutsu and h3-max-recast accept video references but not audio.

Omni Flash takes up to 10 reference images and up to 3 reference videos of at most 3 seconds each, addressed in the prompt as <IMAGE_REF_0> and <VIDEO_REF_0>. A video_url edit source cannot be combined with image or reference fields, so decide up front whether the job is generation from references or an edit of one clip.

Check the catalog before you code against it

Limits change as models are added, and the docs tell you to read capabilities from GET /v1/video-router/models rather than assume one envelope. A one-line check from the shell:

``bash curl -s https://api.sume.com/v1/video-router/models \ -H "Authorization: Bearer $SUME_API_KEY" ``

Note that higgsfield-genjutsu is listed only when its provider is configured, so it may be absent from your response. If a request is rejected for an unsupported parameter, the cause is almost always a per-model limit such as a resolution the id does not offer; the model chooser and the Sora-to-Omni migration guide above cover the usual fixes.