Video generations

Async text-to-video. Submit a job, poll for completion, retrieve the video URL when ready.

Submit a job

POST https://api.ecohash.com/v1/video/generations

Request body

Field	Type	Required	Notes
model	string	yes	Video model ID (e.g. wan21-t2v-1-3b)
prompt	string	yes	Text prompt
num_frames	integer	no	Number of frames. Default 33 (≈ 2s at 16 fps)
fps	integer	no	Playback rate, default 16
num_inference_steps	integer	no	Diffusion steps, default 50
guidance_scale	number	no	How strictly to follow the prompt, default 5.0
seed	integer	no	Deterministic output

Response

{
  "id": "vid_abc123",
  "object": "video.job",
  "status": "queued",
  "created": 1776391234,
  "model": "wan21-t2v-1-3b"
}

Poll for completion

GET https://api.ecohash.com/v1/video/generations/{id}

Response progression

// status: queued → processing → completed (or failed)
{
  "id": "vid_abc123",
  "status": "completed",
  "video_url": "https://ecolink-videos.s3.us-west-1.amazonaws.com/vid_abc123.mp4",
  "duration_seconds": 2.0625,
  "created": 1776391234,
  "completed_at": 1776391312
}

video_url expires after 24 hours. Download and store the bytes if you need them longer.

On failure: status: "failed" with an error field describing the cause.

Timing

Typical end-to-end: 1–3 minutes. Queue wait 30–60s, GPU time 60–120s.

Limits

• Max concurrent jobs per account: 5
• Max num_frames: 81 (≈5s at 16 fps)
• Text-to-video only — no image-to-video input in v1

Errors

See Errors. Common: 401, 402, 404, 429. The async job itself can also fail with status: "failed" mid-generation (prompt too adversarial, internal error) — retry or adjust parameters.