Run One Prompt on Several AI Video Models with One API Key

To run one prompt on several AI video models through Makernice, send the same request with one API key and change only the model field, after removing any field a model doesn't take. A 5-second 720p text-to-video test costs $0.39 with wan-3.0, $0.84 with kling-4.0 and $1.30 with seedance-2.5 (price list v12, November 2, 2026).

Which video models can take the same request?

Pick models that support your mode and duration. For a 5-second text-to-video test at 720p, these five work; GET /v1/models is the live list of each model's modes, durations and fields.

How much does a 5-second test cost on each model?

Prices are from price list v12 (November 2, 2026). The cost calculator's compare view shows your model and up to three others side by side, with the same math. Compare in the calculator →

Step 1: How do you quote the same prompt on each model?

PROMPT="A ceramic mug on a wooden table, steam rising, slow push-in, morning light"
for MODEL in seedance-2.5 kling-4.0 wan-3.0 happy-horse-1.1; do
  curl -s https://api.makernice.com/v1/generations/quote \
    -H "Authorization: Bearer $MAKERNICE_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{"model":"'"$MODEL"'","mode":"t2v","prompt":"'"$PROMPT"'","duration":5,"resolution":"720p"}'
  echo
done

Each response has its own id, amount and expires_at, and quoting holds nothing. For kling-3.0-standard, send the same body without resolution.

Step 2: How do you create and compare the videos?

curl https://api.makernice.com/v1/generations \
  -H "Authorization: Bearer $MAKERNICE_API_KEY" \
  -H "Idempotency-Key: mug-test-kling-4.0" \
  -H "Content-Type: application/json" \
  -d '{
    "quote_id": "qt_...",
    "input": {
      "model": "kling-4.0",
      "mode": "t2v",
      "prompt": "A ceramic mug on a wooden table, steam rising, slow push-in, morning light",
      "duration": 5,
      "resolution": "720p"
    }
  }'

Create each video with its own quote and its own Idempotency-Key, such as the test name plus the model id. An account runs 2 generations at once by default, so a third create returns 429 concurrency_limit_reached with Retry-After until one finishes. Poll each generation, or pass a callback_url and let us POST the results to your server.

Which model differences break a shared request?

A run that fails or times out releases its hold in full, so a model whose run fails costs nothing. Note: if a finished output is blocked by output moderation for a policy violation, the run is still charged. You can contact support for a manual review. To compare specs side by side, see all models or the supported models in the docs.