Pika 2.5 and Pikaframes: Pika's video menu vs Sume's catalog ids

Pika's video menu names six models: Seedance 2.5, MiniMax H3, Wan 3.0, Google Omni Flash, Pika 2.5 and Pikaframes. Sume has catalog ids for the first four's families (seedance-2.5, minimax-h3, wan-3.0, gemini-omni-flash-1.1) and none for Pika 2.5 or Pikaframes. If you want one of Pika's own two models through an API, Sume's OpenAPI enum does not include them.

The Pika list is from the Pika home page, read 2026-10-02. The Sume ids are from the OpenAPI model enum and Video Router.

Which video models does Pika list?

The home page's Video Models menu reads Seedance 2.5, MiniMax H3, Wan 3.0, Google Omni Flash, Pika 2.5 and Pikaframes. Its model carousel also shows Grok Imagine 1.5 and Gemini 1.1 Omni Flash, so the menu is not the whole list. The page does not say how Pika 2.5 or Pikaframes work, so this post does not describe them.

Pika's page names Video Studio as a tool for anything from a short clip to a 30-second multi-shot video, which is a product claim about the app, not about one model.

What do I send on Sume instead?

Pin one of the listed ids, or send model: "sume/auto" and let Sume pick. Sume's docs say the response echoes sume/auto and never discloses which family served the request, so an Auto job cannot stand in for a specific model such as Pika 2.5.

The catalog is the source of truth. A model's supported_resolutions, supported_durations and supported_input_references come from GET /v1/videos/models.

curl https://api.sume.com/v1/videos/models \
  -H "Authorization: Bearer $SUME_API_KEY" \
  | jq -r '.data[].id'

How do I decide whether I need Pika's own models?

If your brief names Pika 2.5 or Pikaframes, use Pika directly and keep that account separate. If the brief is a clip with a given length, resolution and inputs, compare it with the catalog fields for the shared families and pick on those fields. Do not infer one model's behaviour from the other's name.