Grok Imagine Video 1.5 Lite Review: Faster Drafts, but 1.5 Still Wins the Final Cut

We measured Grok Imagine Video 1.5 Lite against Grok Imagine Video 1.5 on seven 480p pairs and six new audio, text, and face checks on October 20, 2026. Lite finished first in 5 of 7 baseline pairs, with a 32.3 second median versus 43.0 seconds. The new face push-in stayed stable for both models.

The short answer: use Lite for drafts, not every final cut

Lite is the better first click when time matters and the brief is simple. Our seven-pair baseline put its median completion at 32.3 seconds, versus 43.0 seconds for Video 1.5. Video 1.5 remains the safer choice when native 1080p, a demanding portrait, or a final client delivery matters more than the first result.

What Grok Imagine Video 1.5 Lite is

Grok Imagine Video 1.5 Lite is xAI's lower-cost video model for text-to-video and image-to-video clips. The official video overview lists 1 to 15 seconds, 480p, 720p, and 1080p output, plus native audio with lip-synced speech. The same page says Lite reaches 1080p by upscaling 720p. Video 1.5 renders native 1080p.

You can try both models from the Grok Imagine overview. The Lite model page is the direct place to start a generation. Vidney handles the model access and shows the credit cost before you run it.

Same-prompt speed test: Lite finished first in 5 of 7 pairs

Lite finished first in five of the seven 480p, 6-second pairs we reused from the October 19, 2026 onboarding check. We sent one prompt to each model at the same time, measured wall-clock completion, and kept every completed result, including the two pairs where Video 1.5 finished first.

The median is the useful number here: 32.3 seconds for Lite and 43.0 seconds for Video 1.5. This is a small production sample, so the result supports a faster-first workflow, not a promise that every Lite clip will finish first.

We then added three new paired checks on October 20, 2026. Each clip was six seconds, and each pair used the same prompt and framing.

The outputs are published as short, compressed review media. Here is the dialogue pair at normal playback length:

The left clip is Lite and the right clip is Video 1.5. Both are fictional adult food-cart vendors. The output files were checked with ffprobe and decoded end to end with ffmpeg before publishing.

Audio and lip-sync check

All six new clips had an AAC audio track. Forced-English Whisper confirmed speech in both dialogue clips and the Lite sign clip. The full sign clip and both face clips had no confirmed speech: no_speech_prob was 0.685, 0.873 and 0.896, with mean levels of -44.8 dB, -36.5 dB and -42.8 dB.

An AAC track does not prove speech. For a silent brief, check the audio before publishing. There is no separate sound switch on these model pages.

Six-second prices as of October 20, 2026

The current Starter calculation is $9 for 300 credits. We recomputed the six-second figures from the live model records while writing this review. These are Vidney credit prices, and public list prices can change.

The table above uses the Vidney credit figures shown before generation, calculated as of October 20, 2026.

The 1080p choice is the clearest tradeoff. Lite is $3.78 for six seconds on Starter and reaches that size by upscaling 720p. Video 1.5 is $1.80 for six seconds and renders native 1080p. If the delivery spec is truly 1080p, the more capable model is also the cheaper choice in this table.

Face push-in: the new pair did not reproduce the earlier Lite drift

The October 19 flower-market push-in sample showed the Lite face drifting away from the still image across three earlier attempts. We ran one new paired push-in on October 20 with the same fictional florist reference. This time the face, hair, and freckles stayed visually stable in both clips, so the earlier failure is a risk signal rather than a universal Lite failure.

Both new face clips are published for inspection:

For a portrait that must match a reference on the first delivery, Video 1.5 still gives us more headroom. For a draft, storyboard, or social variation, Lite is worth trying first and checking frame by frame before approval.

When to pick Lite

Choose Lite when the first useful draft matters more than a guaranteed first-pass match. It suits short social clips, prompt exploration, quick product motion, and image-to-video work where you can afford one review pass. Its 480p six-second price is the same as Video 1.5 on Starter, while the measured median was faster.

Choose Video 1.5 when the brief calls for native 1080p, a close portrait match, or a client-ready result with less time for retries. Its 720p and 1080p credit prices are lower than Lite in the current table, and the new face test does not erase the earlier Lite drift signal.

Method and limits

We reused seven completed Lite and Video 1.5 pairs from October 19, 2026, then ran six new clips on October 20. The baseline used 480p, six-second clips. The new set used three identical prompt pairs: 480p dialogue, 720p sign text, and a 480p image-to-video face push-in. Every result was kept, published in compressed form, checked with ffprobe and decoded end to end with ffmpeg, then reviewed with one-second frame grids. The dialogue and non-dialogue audio were checked with forced-English Whisper.

The sample contains fictional adults only. It is a small review set, not a quality benchmark. We did not score aesthetics with a panel, test every aspect ratio, or claim that the face result generalizes to all portraits. The 1080p Lite output is described as upscaled because that is how the official documentation describes it.

FAQ

Is Grok Imagine Video 1.5 Lite free?

No. Each clip costs credits. New accounts receive 10 to 30 free credits depending on country. As of October 20, 2026, a six-second 480p Lite clip costs 18 credits, so the full 30-credit grant covers one clip; lower tiers may not.

Can Lite make a video from text and from an image?

Yes. The model page supports text-to-video and image-to-video. Our new face check used the same fictional florist still for Lite and Video 1.5.

Is Lite native 1080p?

It can deliver a 1080p-sized file, but the official xAI overview says Lite reaches 1080p by upscaling 720p. Video 1.5 is the better choice when native 1080p detail is part of the brief.

Does Lite generate sound and lip-sync?

All six new clips had an AAC track, but forced-English Whisper confirmed speech only in the two dialogue clips and the Lite sign clip. The full sign clip and both face clips had no confirmed speech, so check audio before publishing.

Does Lite preserve a face from a reference image?

It preserved the face in our new one-shot pair, but an earlier three-attempt flower-market push-in showed drift. Treat portrait continuity as something to inspect, not a guaranteed property.

Is Lite always faster than Video 1.5?

No. Lite finished first in 5 of 7 baseline pairs, with medians of 32.3 seconds versus 43.0 seconds. Two baseline pairs went to Video 1.5 first.

Which model should I try in Vidney?

Start with Grok Imagine Video 1.5 Lite for a quick draft or a simple social cut. Pick Grok Imagine Video 1.5 for native 1080p, a high-stakes portrait, or a final delivery where a retry is expensive.

Verdict

Pick Lite if you want a lower-friction first pass and can review the result before publishing. Pick Video 1.5 if native 1080p, portrait continuity, or a dependable final cut is worth more than the faster first result. The October 20 evidence supports Lite as a fast draft model, not as a universal replacement for 1.5.