Grok Imagine video 1.5 lite $0.02/s vs 1.5 $0.08/s: 60 seconds

xAI's models page (read 2026-10-02) lists grok-imagine-video-1.5-lite at $0.020 per second, grok-imagine-video at $0.050 and grok-imagine-video-1.5 at $0.080. Sixty seconds of output footage is therefore $1.20 on lite, $3.00 on the original model and $4.80 on 1.5, before any retries.

The three video rows

The page I read shows one rate per model and no resolution tiers, so these numbers apply per output second as listed. The 8-second column is a common clip length and is simply the rate times 8.

Retries are the real multiplier

A price per second tells you the cost of a keeper, not of a finished video. If you keep one take in three, your effective rate is three times the list. That is why a 4-times price gap between lite and 1.5 can reverse: if lite needs twice the takes to get an acceptable result, the real gap is 2 times, and if it needs four times the takes, there is no saving. Count kept seconds, not generated seconds, when you compare.

Images on the same page

The same page lists image rows: grok-imagine-image at $0.02 per image, grok-imagine-image-2.0 at $0.04 and grok-imagine-image-quality at $0.05. For a storyboard of 12 frames, that is $0.24, $0.48 or $0.60, which is small next to the video cost, so spend the retries on stills before you commit to video seconds.

What to check on Sume

Sume's catalog includes a Grok Imagine video id, grok-imagine-video-1.5, in its Video Router. I did not find a lite id in the Sume docs I read, so confirm with GET /v1/video-router/models before you plan around it. Do not assume xAI's list price is Sume's list price: Sume prices from the provider route it uses.

The documented rule, in the video docs, is that a job reserves at provider list times 1.25 on submit and reports usage.cost as the billable amount. Whatever the catalog shows for the model is what to multiply by your kept seconds.