Grok Imagine API: the cheapest way to call it

Video: Grok Imagine Video 1.5

Grok Imagine Video 1.5 does text to video, image to video, first and last frames and reference to video, up to 1080p, with sound. A clip runs 1 to 15 seconds in 16:9, 9:16 or 1:1, and you can send up to 4 reference images.

Prices are dollars per second of video. A 10 second clip at 1080p is $1.30 on VidLab and $3.92 at Higgsfield's Plus rate.

Image: Grok Imagine Image 2.0

Grok Imagine Image 2.0 does text to image and image editing, up to 2K, with up to 3 reference images and up to 4 images per request.

Higgsfield sells this model by quality tier. The figures here are for its lowest tier, which is the cheapest it offers, so the comparison is against Higgsfield's best price for the model. VidLab also lists 1.5K at $0.02, a size Higgsfield does not offer, so there is nothing to compare it with.

Every Grok Imagine model

To pick between the video models, use Grok Imagine Video or the preview for drafts, since they have the lowest price per second, and move to 1.5 when you need first and last frames or a reference video.

Call it

curl -s -X POST https://vidlab.ai/v1/generations \
  -H "Authorization: Bearer $VIDLAB_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $(uuidgen)" \
  -d '{"model":"grok-imagine-video-1.5","input":{"prompt":"A paper boat drifting down a rain-wet street at dusk","resolution":"720p","duration":5,"aspect_ratio":"9:16"},"max_price_usd":0.35}'

The answer is 202 with a job id. Poll GET /v1/generations/{id} every 5 to 10 seconds until it is succeeded, then download the file from outputs. Run POST /v1/estimate with the same body first if you want the price before you start. For the image, swap the model for grok-imagine-image-2.0 and send resolution as 1k, 1.5k or 2k.