Grok Imagine pricing:
what every Grok surface costs on Playcut
Three different products ship under the Grok Imagine name, on two different billing units, at three different rates. That is why "Grok Imagine is cheap" is not a budget. This page gives the credit rate for all three surfaces Playcut runs — the image model, classic video, and 1.5.
The rate below is what your balance is debited. Credits are held when you submit and reversed automatically if the job fails or is cancelled.
Rates in credits · you pick the model, every time · every paid plan reaches all three
Grok Imagine pricing at a glance
Three products share the Grok Imagine name, and they are priced on two different units. The image model bills per finished image; both video models bill per second of output, so a 10-second clip costs ten times the listed rate.
Which Grok surface should you pick?
These are separate models that happen to share a name, not tiers of one product. You choose the model on every generation from the studio's picker. Each capability has a default — Nano Banana Pro for images, Veo 3.1 for video — and Grok is never one of them, so nothing routes you here automatically.
Grok Imagine Image — 10 credits, the cheapest image on Playcut
At 10 credits it is the lowest image rate in our catalog. For comparison: Nano Banana 2 runs 20 at 512px, 34 at 1K / 2K and 43 at 4K; Nano Banana Pro runs 67 at 1K / 2K and 84 at 4K.
Shape is the other reason to reach for it. Six of its 14 ratios — 2:1, 1:2, 19.5:9, 9:19.5, 20:9 and 9:20 — exist on no other image model we run, which covers banner slots and full-bleed phone screens without a crop step. Its auto mode hands the ratio choice to the model, useful when you are batching and do not want to decide 40 times.
What it costs you: a 2K ceiling with no 4K tier, a maximum of two reference images, and no text rendering. Reference-to-image is 20 credits, double the text-to-image rate.
Grok Imagine Video classic — the cheapest 720p second
At 20 credits a second it is the lowest 720p rate of our seven video models — Seedance 2.0 is 21, Grok Imagine 1.5 is 32, Gemini Omni Flash is 40, Wan 2.7 is 40, HappyHorse 1.1 is 56, Veo 3.1 is 160. It runs four modes: text-to-video, image-to-video, reference-to-video with up to 7 references, and extension.
Shape is where it stands alone. Both Grok video models accept 3:2 and 2:3, and none of the other five video models accepts either — Veo takes 16:9 and 9:16, and Seedance, Omni, HappyHorse and Wan all carry lists without those two. Text-to-video runs 1 to 15 seconds; the other three modes cap at 10.
Grok Imagine 1.5 — animate only, at 1.6× the rate
32 credits a second buys a sharper image-to-video pass and nothing else. Text-to-video, reference images, extension and interpolation are all rejected before any credits are held. It is a preview model; classic is stable.
The extra 12 credits a second is worth it when you already have a still you like and the animation quality is the whole job. If you are starting from a prompt, or you need references, extension or a second opinion cheaply, classic is simply the better buy.
What a Grok Imagine job actually costs
Per-second rates are hard to feel. Here is the arithmetic for jobs you can actually place in the studio — every clip length below sits on the duration ladder — with a Veo 3.1 line at the bottom for contrast.
Draft cheap, finish expensive
The last two rows are the whole argument. Eight seconds costs 8× as much on the flagship as on Grok, so the sane order is to answer "does this shot idea read at all?" at 20 credits a second and only then re-run the survivor at 160 credits a second.
Prompts port between models with light editing, and everything lands in the same workspace on the same balance, so promoting a winner costs credits and nothing else. The same pattern works on stills: explore on Grok at 10 credits, ship on Nano Banana Pro.
How to run Grok Imagine
You need a Playcut account, an active plan and a credit balance. There is no xAI key to supply and no per-model unlock: every paid tier, Hobby through Agency, reaches all three surfaces. A workspace with no subscription falls back to image-only tool access, so the video models need a plan behind them, not just credits.
In the studio
Open the model picker on the generation you are about to run and choose the Grok surface you want. The pick is per generation, so you can draft on Grok and finish on Veo inside one session.
One thing to expect: the duration dial is a fixed ladder — 1, 3, 5, 8, 10, 12, 15 seconds — filtered by the ceiling for the mode you are in. Six and seven seconds are not rungs, so cut your edit to the ladder rather than around it. The REST tools API accepts any whole number from 1 to 15 if you need one.
Over the API and MCP
The split is uneven and worth knowing before you write the integration. All six video tools in the registry take the model as a free-form string, so nothing in the schema blocks a Grok id. What the service accepts is narrower: classic on four modes, 1.5 on generate-video-from-image alone, and neither on interpolation or video edit. The two image tools enumerate the Gemini models only, which makes Grok Imagine Image studio-only.
# The 5-second cutdown from the table above — 100 credits, at 2:3, # a ratio none of the other six video models will take. curl -X POST https://api.playcut.ai/api/v1/tools/generate-video-from-text/execute \ -H "Authorization: Bearer $PLAYCUT_API_KEY" \ -H "Content-Type: application/json" \ -d '{"workerId": "…", "model": "grok-imagine-video", "prompt": "A ceramic mug on a windowsill, steam rising, morning light creeping across the counter", "aspectRatio": "2:3", "resolution": "720p", "durationSeconds": 5}'
The response is a task id; poll it and the credits settle on their own. Swap durationSeconds to 15 for the hero cut and the same call bills 300 credits. Field-level specs, the other mode calls and error handling live in the Grok Imagine video model reference.
Honest limits — what Grok Imagine will not do
Every item here is enforced in our own code or by xAI, and most of them are checked before a single credit is held. Read this section before you plan a deliverable around a Grok rate.
- Both video models stop at 720p
- There is no 1080p and no 4K anywhere in the Grok family. A request for either is rejected with a 400 before any credits are held — nothing is silently downgraded to a resolution you did not ask for.
- Resolution is not a cost lever on Grok
- Our price map is keyed on the model, not the pixel count, so a 480p second and a 720p second both bill 20 credits on classic and 32 on 1.5. Seconds are the only lever that moves a Grok invoice — cut a beat, not the resolution.
- Neither video model generates audio
- Every Grok clip comes back silent. They are the two exceptions among our seven video models — Veo 3.1, Seedance 2.0, Gemini Omni Flash, HappyHorse 1.1 and Wan 2.7 all produce audio. Budget a separate audio pass.
- Grok Imagine 1.5 does image-to-video and nothing else
- Text-to-video, reference images, extension and interpolation are all hard 400s before the hold. It costs 12 credits a second more than classic for strictly fewer modes.
- No Grok model does interpolation or video edit
- Interpolation is Veo 3.1 and Seedance 2.0. Prompt-driven video edit is Wan 2.7 and HappyHorse 1.1. Neither flow accepts a Grok model, and asking for one returns an error rather than a substitution.
- 15 seconds is a text-to-video privilege
- Start from a still or a reference and the ceiling drops to 10 seconds; an over-length request is refused rather than trimmed. So a 15-second Grok hero has to be a prompt-only shot, or a 10-second animate plus a classic extension on top.
- Grok Imagine Image is studio-only
- The two image tools validate the model field against a Gemini-only list, and the tool description says so outright. You cannot script Grok Imagine Image over MCP or the REST tools API today. Both Grok video models are reachable.
- Two reference images on the image model, not five
- xAI's own pages disagree with each other on this. A live probe found their parser bounds the array at three and that exactly three trips a bug on their side, so we enforce two before the hold rather than let you pay for a failure.
- Budget edits separately from originals
- An edit bills 20 credits, an original 10. So 40 previews plus 10 fixes is not 50 × 10 — it is 600. On the Gemini models both flows cost the same, which narrows Grok's advantage on edit-heavy work.
- Not the model for text inside an image
- Grok Imagine Image is marked as not doing text rendering in our feature matrix; both Nano Banana models are marked as doing it. If the render needs a legible sign, label or packaging line, this is the wrong 10 credits.
- A scripted extension is always exactly 8 seconds
- The public extension tool pins duration to a literal 8 and resolution to a literal 720p, so every extension you script adds 8 seconds and 160 credits on classic — no more, no less. The studio picker still gives you the shorter rungs.
Why one brand name covers three different products
Grok Imagine is a family, not a model. xAI ships an image generator, a video generator, and a second-generation video model that only animates stills, and all three carry the same name in marketing copy. On Playcut they are three separate picker entries with three different rates and three very different capability sets.
That matters for budgeting more than it matters for taste. The image model bills per finished image; both video models bill per second of output. A team that reads "Grok Imagine is cheap" and plans a 15-second hero around it will discover the 720p ceiling and the silence at the worst possible moment.
The release order explains the shape. The image model arrived in November 2025, classic video in December 2025, and 1.5 in May 2026. Classic is marked stable in our catalog; 1.5 is still marked preview, which is part of why it carries the narrower mode list.
Where Grok genuinely wins, and where the win stops
On price, the image model is a clear first. At 10 credits it undercuts every other image surface we run, and the gap is not marginal — Nano Banana Pro costs 6.7× as much per render. If your job is to look at a hundred compositions and keep three, this is the model that makes that affordable.
On shape, both video models are the only route to 3:2 and 2:3 output on Playcut. That is a genuinely exclusive capability rather than a marketing line: Veo does 16:9 and 9:16, Seedance and HappyHorse each carry their own longer lists, and none of them include those two ratios.
On video price, the honest claim is narrower than the obvious one. Classic Grok is the cheapest 720p second at 20 credits, but Seedance 2.0 at 480p is 10 credits a second — half the rate. The qualifier is load-bearing, and anyone telling you Grok is the cheapest video model in the catalog has skipped a row.
The win stops at resolution and sound. Nothing in the Grok family reaches 1080p video or 4K anything, and neither video model produces audio. Those are not settings you can turn on; they are the shape of what xAI ships, and we surface the refusal rather than quietly rerouting your job to a model that costs 8× as much.
How the credit price works here
One balance covers everything on Playcut — images, video, music, voice, actor performances. There is no per-model subscription, and the paid tiers differ on how many credits arrive each month and how many seats you get, not on which models you may touch. A Hobby account can run Veo 3.1; it just burns the allowance faster than the same work on Grok.
Your balance moves once, at submit: one atomic decrement plus a ledger entry, before anything reaches a worker. Should the run end failed or cancelled, that entry is reversed and the credits come back. Bad parameter combinations never reach the hold at all — an unsupported resolution, an off-list ratio or an over-length duration returns a 400 first, so a typo costs you a retry and nothing else.
Every figure on this page is in credits, so you can total a campaign against your balance before committing to it. The pricing page shows how many credits each plan includes.
You pick the model. Every time.
There is no classifier reading your prompt and deciding which model gets your credits. Images default to Nano Banana Pro and video defaults to Veo 3.1; change nothing and those are what render. Grok is never a default, so a Grok job is always a Grok job you asked for.
We are explicit about this because an automatic router is a pricing problem wearing a convenience costume. If a hidden system could move a job from a 20-credit model to a 160-credit one, no campaign budget on this page would survive contact with reality. Fixed defaults plus a visible picker means the rate you read is the rate you are billed.
The trade-off is that you have to know the models. That is what the full model comparison is for, and why this page spends more words on refusals than on features. Video workflows in production terms live on the AI video generator page; the image flows are covered on the AI image generator page.
Common questions
How much does Grok Imagine cost on Playcut? +
Four rates. An image is 10 credits; editing one with a reference is 20 credits; classic Grok Imagine video is 20 credits per second; Grok Imagine 1.5 is 32 credits per second. That is what your balance is debited, with no separate per-model fee on top — every paid plan, Hobby through Agency, reaches all three surfaces.
Is Grok Imagine the cheapest model on Playcut? +
For images, yes — 10 credits is the lowest image rate in the catalog. For video, it is the cheapest 720p second at 20 credits, but not the cheapest second outright: Seedance 2.0 at 480p is 10 credits per second. Seedance carries its own catch, though — it rejects input images containing human faces, so for anything with a person in the source frame Grok is usually the cheapest usable option.
What is the difference between Grok Imagine classic and 1.5? +
Classic runs four modes — text-to-video, image-to-video, reference-to-video and extension — at 20 credits a second, and takes up to 7 reference images. Grok Imagine 1.5 runs image-to-video only, at 32 credits a second, and hard-rejects text-to-video, reference images, extension and interpolation. Pick 1.5 when you already have the still and want the better animate. Pick classic for everything else, including any job that starts from text.
Can I use Grok Imagine through the API or an MCP client? +
The video models yes, the image model no. Classic takes model: grok-imagine-video on generate-video-from-text, generate-video-from-image, generate-video-from-reference and generate-video-extension. Grok Imagine 1.5 is accepted on generate-video-from-image only — the other three reject it with a 400 before any credits are held. The two image tools validate the model field against a Gemini-only enum, so Grok Imagine Image runs in the studio only today.
Does Grok Imagine generate sound? +
No. Both Grok video models return silent clips, and there is no flag that changes that. They are the only two of our seven video models without audio generation — Veo 3.1, Seedance 2.0, Gemini Omni Flash, HappyHorse 1.1 and Wan 2.7 all produce it. If the shot needs a track or a voice, generate the audio separately and lay it under the clip.
Does Playcut pick Grok for me automatically? +
No. You pick the model on every generation. Each capability ships a default so the picker is optional — Nano Banana Pro for images, Veo 3.1 for video — and Grok is never one of those defaults. You get Grok only when you choose it, which also means a Grok rate can never turn into a Veo rate on your invoice.
What happens if a Grok job fails? +
Your balance moves once, at submit — one atomic decrement before anything reaches a worker — and it reverses on its own when a run ends failed or cancelled. Invalid combinations never get that far: an unsupported resolution, an off-list aspect ratio or an over-length duration is refused with a 400 ahead of the hold, so a mistyped parameter costs you nothing.
Related
- Every AI model Playcut runs, comparedAll 13 surfaces, one price list.
- AI video generatorThe five video types in production terms.
- AI image generatorThe four image flows and the three image models.
- Plans and credit allowancesWhat each tier includes per month.
- Grok Imagine Image reference (docs)Full specs and the complete aspect-ratio table.
- Grok Imagine video reference (docs)Mode semantics and the full set of API calls.
Ready to make something cinematic?
Pick a plan. Hobby starts with a 7-day trial — cancel within 7 days at no charge.