MixCut generation

Send type: video, model: mixcut, a list of public source video URLs in input.urls and a natural-language input.requirement. The submit probes every source, reserves credits for the measured input plus the 10-minute output cap, and returns HTTP 202 with job.id and an initial queued/running state, without waiting for rendering. Poll GET /v1/generations/{id}; state moves queued → running → done/failed, step moves analyze → generate → render → done, and a finished task carries video_url.

Request

curl -X POST https://api.anycap.io/v1/generations \  -H "Authorization: Bearer $ANYCAP_API_KEY" \  -H "Content-Type: application/json" \  -H "Idempotency-Key: 6d2f9a1c-7b4e-4a8d-9f3c-2e5b8d1a7c60" \  -d '{"type":"video","model":"mixcut","input":{"urls":["https://mixcut-files.tos-cn-beijing.volces.com/ecommerce/8.mp4","https://mixcut-files.tos-cn-beijing.volces.com/ecommerce/9.mp4"],"requirement":"做一个22秒的产品种草短视频"}}'

Response

{  "type": "video",  "job": {    "id": "6d2f9a1c-7b4e-4a8d-9f3c-2e5b8d1a7c60",    "state": "queued",    "status": "queued",    "step": "analyze",    "urls": [      "https://mixcut-files.tos-cn-beijing.volces.com/ecommerce/8.mp4",      "https://mixcut-files.tos-cn-beijing.volces.com/ecommerce/9.mp4"    ],    "requirement": "做一个22秒的产品种草短视频",    "input_seconds": 22,    "reserved_microcredits": 1415000000,    "created_at": "2026-11-02T08:00:00Z",    "updated_at": "2026-11-02T08:00:00Z"  }}

The response wraps the task in job. input_seconds and output_seconds are the measured durations used for billing; reserved_microcredits is the hold created at submit; charged_microcredits appears after settlement. A failed task carries error and releases the hold.

Query progress and result

curl "https://api.anycap.io/v1/generations/$TASK_ID" -H "Authorization: Bearer $ANYCAP_API_KEY"

Set TASK_ID to job.id from the create response (the Anycap UUID). Query every 3 seconds, waiting for each response before the next query. Keep polling queued/running. On done read job.video_url; on failed read job.error. video_url, output_seconds and completed_at are omitted until available. A temporary query error does not mean generation failed: retry GET, without submitting another POST. If a nonterminal job carries error, its submission outcome needs reconciliation; contact support and do not resubmit.

{  "type": "video",  "job": {    "id": "6d2f9a1c-7b4e-4a8d-9f3c-2e5b8d1a7c60",    "state": "done",    "status": "done",    "step": "done",    "video_url": "https://cdn.example.com/output.mp4",    "output_seconds": 22  }}

Errors