Grok Imagine on HiAPI is billed per call, not by subscription. Image endpoints charge per generated image, from $0.03, and video endpoints charge per second of output, from $0.015 per second. You pay only for successful results, there is no monthly fee and no minimum spend, and new accounts start with 200 free credits.
The whole grok-imagine family is live on HiAPI. Open the text-to-image model page to generate right away, or browse every model on the HiAPI pricing page. Prices below were verified against HiAPI's live pricing on October 17, 2026; re-check the pricing page before you budget, as rates change.
How Grok Imagine billing works on HiAPI
Billing is pay-as-you-go and metered per result:
- Images bill per image. One successful image is one charge, whatever the prompt length or aspect ratio.
- Videos bill per second. Your charge is the per-second rate multiplied by the clip duration you request.
- Failed tasks are refunded. Billing counts successful outputs only, so a task that errors out does not cost you anything.
- No subscription, no minimum. You are not committing to a monthly plan; spend tracks output one call at a time.
- Credits convert at 2,000 credits = $1. The model pages show prices in credits; divide by 2,000 to read them in dollars. The 200 sign-up credits are worth $0.10.
Every grok-imagine endpoint runs on HiAPI's unified asynchronous task API: you POST /v1/tasks, then poll with the returned taskId or receive the result on a callback.url. The request and response shapes are in the API docs.
What does each Grok Imagine endpoint cost?
Grok Imagine is a family, and the price depends on which endpoint you call. Image endpoints are priced per image:
| Endpoint | Model ID | Price per image | Resolution |
|---|---|---|---|
| Text-to-image | grok-imagine/text-to-image | $0.03 | 1K and 2K, same price |
| Image 2.0 text-to-image | grok-imagine-image-2.0/text-to-image | $0.052 | 1K and 2K, same price |
| Quality text-to-image | grok-imagine-quality/text-to-image | $0.07 (1K) / $0.10 (2K) | tiered by resolution |
| Image-to-image | grok-imagine/image-to-image | $0.03 + $0.003 per reference | 1-3 reference images |
| Image 2.0 image-to-image | grok-imagine-image-2.0/image-to-image | $0.065 | 1K and 2K, same price |
| Quality image-to-image | grok-imagine-quality/image-to-image | $0.08 (1K) / $0.10 (2K) + $0.015 per reference | 1-3 reference images |
Two of those rows carry a small reference add-on. grok-imagine/image-to-image is $0.03 per image plus $0.003 for each reference image you pass, so a one-reference edit is $0.033 and a three-reference blend is $0.039. grok-imagine-quality/image-to-image is $0.08 at 1K or $0.10 at 2K, plus $0.015 per reference image. Everything else in the image table is a single flat price.
Video endpoints are priced per second, and the only thing that moves the rate is resolution:
| Endpoint | Model ID | 480p | 720p | Duration |
|---|---|---|---|---|
| Text-to-video | grok-imagine/text-to-video | $0.015/s | $0.0282/s | 6-30s |
| Image-to-video | grok-imagine/image-to-video | $0.015/s | $0.0282/s | 6-30s |
| 1.5 image-to-video | grok-imagine-1.5/image-to-video | $0.015/s | $0.0282/s | 1-15s |
Because video bills per second, a short clip is cheap and a long one scales linearly. A 6-second clip at 480p is 6 x $0.015 = $0.09; the same clip at 720p is 6 x $0.0282 = $0.169. A full 30-second 480p clip is $0.45, and at 720p it is about $0.85.
Two endpoints have their own deep-dive pricing pages if you need the full breakdown for a specific model: Grok Imagine 1.5 video pricing and Grok Image 2.0 pricing. This page is the family-wide view.
The image below was generated with grok-imagine/text-to-image, the family's $0.03 entry endpoint, at 2K. At one flat price for both 1K and 2K, there is no reason to render finals at the lower resolution:
matte ceramic pour-over coffee dripper on a pure white seamless background, studio softbox lighting, centered, no shadow gradient beyond a soft contact shadow, no text, no watermark, no props, e-commerce product photography, sharp focus
Generate with this prompt → (set the resolution to 2K)
What will a month of Grok Imagine cost?
Your monthly bill is calls multiplied by the unit price, nothing more. For images, pick the column that matches the endpoint you will actually run:
| Images per month | Text-to-image ($0.03) | Image 2.0 ($0.052) | Quality 2K ($0.10) |
|---|---|---|---|
| 100 | $3 | $5.20 | $10 |
| 1,000 | $30 | $52 | $100 |
| 10,000 | $300 | $520 | $1,000 |
For video, multiply the number of clips by the per-clip cost. These figures assume 6-second clips:
| 6-second clips per month | 480p ($0.09) | 720p ($0.169) |
|---|---|---|
| 100 | $9 | $16.92 |
| 1,000 | $90 | $169.20 |
A $10 trial budget buys about 333 text-to-image generations at $0.03, about 192 at $0.052, or 100 Quality renders at 2K. The 200 sign-up credits cover three base images outright, or one Quality 2K image.
How to estimate and control cost at scale
- Match the endpoint to the job. Base
grok-imagine/text-to-imageat $0.03 is for drafts, variations and high volume; it is about 57% cheaper than the Quality tier at 1K. Reservegrok-imagine-qualityfor the hero frames where extra detail earns its keep. - Render image finals at 2K on the flat-priced endpoints. Base text-to-image and Image 2.0 charge the same at 1K and 2K, so there is no saving in rendering small and no second upscale pass to pay for.
- Estimate video before you submit. Cost is seconds x per-second rate, so a 6-second draft at 480p ($0.09) is roughly 47% cheaper per second than 720p. Draft at 480p, finish the keepers at 720p.
- Keep reference counts low on image-to-image. Each reference adds $0.003 on the base endpoint and $0.015 on Quality. Pass only the references the edit needs.
- Submit each task once. Poll with the
taskIdor wait for thecallback.url; every repeated POST creates and bills a new task. - Price-test for free first. Failed tasks are refunded and the 200 sign-up credits let you confirm quality on each endpoint before you scale.
Pay-per-call vs a subscription
HiAPI bills Grok Imagine the way programmatic work wants to be billed: a fixed amount for each successful image or each second of video, with no recurring commitment. Cost scales linearly and predictably with output, so 1,000 images is exactly 1,000 times the per-image price and you can put a number on a job before you run it.
A flat subscription works differently. You pay one recurring fee for interactive, manual use, which is convenient for one-off creation but gives you no per-asset rate to forecast against and no programmatic access for batch or automated pipelines. We do not quote a first-party API rate to compare against here, because we cannot verify one at the time of writing; compare on billing structure, not on an unverified number. For automation, batch jobs and anything you need to meter and budget, per-call pricing is the model that keeps every run countable.
FAQ
Is Grok Imagine a subscription on HiAPI? No. It is pay-as-you-go. You are charged per successful image or per second of video, with no monthly plan and no minimum spend.
How much does one Grok Imagine image cost?
From $0.03 for grok-imagine/text-to-image up to $0.10 for the Quality tier at 2K. Image 2.0 text-to-image sits in between at $0.052.
How much does a Grok Imagine video cost? It is priced per second: $0.015/s at 480p and $0.0282/s at 720p. A 6-second 480p clip is about $0.09, and a 30-second 480p clip is $0.45.
Does 2K always cost more than 1K? No, it depends on the endpoint. Base text-to-image and Image 2.0 charge the same for 1K and 2K; only the Quality tier charges more at 2K ($0.10 vs $0.07).
Am I charged for failed generations? No. Billing counts successful outputs only, and failed tasks are refunded automatically.
Can I try it for free? Yes. New accounts get 200 credits, worth $0.10, which covers three base text-to-image runs or one Quality 2K image.
Start generating with Grok Imagine
Open the grok-imagine text-to-image model page to run your first image on the $0.03 endpoint, or check the HiAPI pricing page to compare it with every other image and video model before you commit a budget.
