Seedance 2.5 vs 2.0 API: models and pricing

TTAPI lists four Seedance video API IDs. ByteDance positions Seedance 2.5 around longer audio-video storytelling, precise reference control, and editing. Its Seedance 2.0 overview emphasizes text, image, audio, and video references with synchronized sound. The Fast and Mini entries are separate 2.0-family IDs in the TTAPI catalog; treat their names as product tiers, not measured quality or latency results.

Start with the task: a story with controlled camera and references, a multimodal clip, or a lower-cost draft. Keep the prompt, input media, duration, resolution, and acceptance checks fixed when you compare IDs. A provider showcase can illustrate a look, but only a test on your own brief can establish which output is useful to you.

ByteDance describes Seedance 2.5 as built for narrative clips, stronger reference interpretation, and more reliable editing. Its public miniature example uses a handmade felt world, a small animal, soft light, and deliberate stop-motion movement. Watch for whether the motion and material texture stay coherent rather than judging a still frame alone.

The provider's product page discusses clips up to 30 seconds. The current TTAPI text-to-video request schema in this catalog lists duration values of 4–15 seconds or -1 for smart selection, so follow the endpoint contract for a TTAPI request instead of copying a showcase duration blindly.

Seedance 2.0 combines visual generation with synchronized audio and accepts more than a text prompt in its multimodal workflows. The published pizza-shop scene shows why motion, sound, and the relationship between foreground and background belong in the review brief.

On TTAPI, text-to-video, image-to-video, omni-reference, edit, and extend are distinct request shapes even when they share a submit endpoint. Use the reference for the chosen workflow before adding image, video, or audio inputs. The edit and extend descriptions call out Seedance 2.5 specifically, so confirm support before applying them to a 2.0-family ID.

The TTAPI text-to-video route requires model and content; each content item in this workflow has type=text and a text prompt. The example below requests Seedance 2.5 at 720p, 16:9, and 10 seconds, with generate_audio=true. Those values fit the current schema. The ratio must be explicit when supplied: adaptive is not accepted by the text-to-video route. Keep TTAPI credentials on the server and confirm the current fields before shipping.

A successful submit returns an id for an asynchronous task, not a ready MP4. Persist that id before leaving the request flow. The fetch endpoint uses POST with the id in the URL path; its status values are queued, running, succeeded, failed, cancelled, and expired. TTAPI's current Seedance reference says task retrieval is available for seven days and generated video URLs are valid for 24 hours, so save approved files promptly.

curl --request POST \
  --url 'https://api.ttapi.io/sd/api/v3/contents/generations/tasks' \
  --header "TT-API-KEY: $TTAPI_KEY" \
  --header 'Content-Type: application/json' \
  --data '{
  "model": "doubao-seedance-2-5-260628",
  "content": [
    {
      "type": "text",
      "text": "A handmade felt rabbit crosses a miniature meadow at sunrise. Slow camera push, visible wool fibers, gentle frame-by-frame motion."
    }
  ],
  "ratio": "16:9",
  "duration": 10,
  "resolution": "720p",
  "generate_audio": true
}'

• Use a documented model ID.
• Store the task ID before leaving the request flow.
• Review motion and audio before publishing the clip.
• Check current TTAPI pricing for the chosen model and workflow.

Use the id returned by the submit call to request the same task. A succeeded response places the generated file at content.video_url; when return_last_frame was set on submission, content.last_frame_url may also be present. Keep waiting only while the status is queued or running, and handle failed, cancelled, or expired without creating another paid job automatically.

The example below is a single status check. Replace $TASK_ID with the returned id in your server environment. Download a video you intend to keep because the returned video URL is time limited.

curl --request POST \
  --url "https://api.ttapi.io/sd/api/v3/contents/generations/tasks/$TASK_ID" \
  --header "TT-API-KEY: $TTAPI_KEY"