Audio-driven talking/singing video from a single image plus audio.
480p $0.35 · 720p $0.70 per request
video model · per request, by resolution (480p or 720p) · one API key, one prepaid balance, no deployment.
Get an API keycurl -X POST https://gpu.swarmdo.com/api/managed/infinitetalk/runsync \
-H "Authorization: Bearer $SWARMDO_API_KEY" -H "content-type: application/json" \
-d '{"input":{"prompt":"a calm greeting","image":"https://example.com/face.png","audio":"https://example.com/line.mp3","size":"480p"}}'
{"id":"…","status":"COMPLETED","output":{"video_url":"https://…/out.mp4"}} ← x-swarmdo-charged-cents header carries the charge
| Tiers | 480p: $0.35 · 720p: $0.70 — the request must name one |
| Long jobs | A job that outlives the call returns an id; poll GET /api/managed/infinitetalk/status/{jobId}. The bill settles against what was actually generated. |
| Output retention | The URL in the reply stays downloadable for 7d; copy the file to your own storage before then. |
| Billing | per request, by resolution (480p or 720p); prepaid credit, charged only when the call succeeds; the charge is returned in the x-swarmdo-charged-cents header |
All hosted models: /models · full reference: /docs · SDK: npm i @swarmdo/gpu-sdk · MCP: npx -y @swarmdo/gpu-mcp