Wan 2.6 API

Wan 2.6 Image to video starts from the frame you send in image_url and returns exactly 5, 10 or 15 seconds at 720p or 1080p, with sound. The frame fixes the subject, wardrobe, set, light direction and the proportions of the clip. The prompt is optional and holds up to 1,500 characters: one action for a single take, or numbered shots for a multi-shot edit.

Choose a Wan 2.6 mode by starting material

The frame decides shape and appearance

There is no aspect ratio field: the clip keeps the proportions of the image you send, so a 9:16 delivery starts with a 9:16 crop. The frame can be JPEG, PNG without transparency, BMP or WEBP, 240 to 8,000 pixels on each side and no larger than 20 MB. Everything visible in it is already decided, which leaves the prompt free to describe motion.

Coverage cut from one frame

With shot_type set to multi, the model can cut between angles while keeping the look of the opening frame, so a fifteen-second clip can carry an establishing view, a detail and a payoff instead of one drifting camera move. A multi request always runs with prompt expansion on. Keep single for product turns, talking heads and anything that must stay continuous, because multi breaks continuity by design.

Sound generated or supplied

The clip returns with audio generated in the same pass as the picture. To make speech or musical beats land where you expect, send audio_url: WAV or MP3, 3 to 30 seconds, up to 15 MB. A longer track is cut to the clip length; with a shorter one, the rest of the clip is silent.

A product macro that must not jump

The serum example animates a still of a dropper bottle for five seconds at 1080p with a fixed seed. A single drop falls from the dropper tip, rings spread across the liquid and the light band on the glass brightens slightly, all under locked macro framing with a soft chime of glass. It stays a single take because a product shot should not cut between angles.

A short interior story from one still

The evening apartment example runs 15 seconds at 720p with shot_type set to multi and prompt expansion on. One interior frame supplies three shots: the lamp flickers as city lights blink on outside, a folded blanket slides off the armchair, and the front door swings open to spill hallway light across the floorboards, under quiet room tone and distant traffic.

A character frame speaking your own line

Pair a character still with a recorded voice line in audio_url so the speech lands where you expect rather than as generic motion. Talking heads belong on single, and the full length is worth checking: when the track is shorter than the clip, the remainder plays silent.

# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
  -H "Authorization: Bearer $SPICY_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
  -d '{
  "model": "alibaba/wan-2.6/image-to-video",
  "input": {
    "duration_seconds": 5,
    "image_url": "https://example.com/reference.png",
    "resolution": "720p",
    "shot_type": "single"
  }
}'

# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
  -H "Authorization: Bearer $SPICY_API_KEY"

See the API reference for authentication, the complete schema, polling, and error handling.