Cheapest Image and Video Generation APIs: Live Prices on HiAPI

TL;DR

• The cheapest images on HiAPI today: FLUX.2 Klein 4B from $0.00036 (0.25 MP) or $0.00143 at 1 MP, FLUX.1 Schnell at $0.005, Z-Image at $0.008. The cheapest video: Grok Imagine at $0.015 per second at 480p.
• The number that decides your bill is cost per image you keep: list price × tries to get an acceptable result. A $0.005 model you run five times costs more than a $0.02 model that gets it right once.
• The cheapest workflow is to draft on a low-cost model and render only the keeper on a better one. On HiAPI that's a one-word change to model, with one key and pay-as-you-go billing. New accounts get 200 free credits.

Cheapest image models on HiAPI

Live HiAPI prices, checked October 20, 2026. Failed tasks aren't charged. Prices change; the pricing page is always current.

The table leaves out tools that aren't general image generators, such as background removers, and models that are listed but not yet open.

Cheapest video models on HiAPI

Same date and source as above.

Video is where the per-second rate matters most: a 10-second clip costs twice a 5-second one. Draft at the lowest resolution and the shortest length, and go up only for the version you keep. Kling 2.5 Turbo is the cheapest way to get roughly 1080p output; details are in Kling 2.5 Turbo API pricing.

Cost per image you keep

The rate in a price table is the cost per image generated. Your budget depends on the cost per image you keep:

cost per kept image = price per image × tries to an acceptable result

Only your own prompts can tell you the "tries" column. Run 20 representative prompts on two or three candidates, count how many came out usable, and divide. Things that push the real cost up:

• Retries. Every image you throw away was billed.
• Text inside the image. Cheap models misspell more. FLUX.1 Schnell invents garbled lettering on objects unless you add "no text, no logos".
• Resolution. Don't draft at 4K. Most models charge more for bigger images (Seedream 4.5 and Seedream 5.0 Lite are exceptions).
• Edits. If you'll need to change the image afterwards, a model that takes reference images saves you regenerating from scratch.

FLUX.1 Schnell: real outputs at $0.005

FLUX.1 Schnell is the cheap, fast member of the FLUX family. It takes prompt and aspect_ratio (1:1, 4:3, 3:4, 16:9, 9:16; default 4:3). Each of these took about 26–32 seconds end to end.

A minimalist still-life arrangement on a cream-colored paper backdrop: three vintage instant cameras stacked at slightly different angles, soft morning side-light casting long pastel shadows, a few scattered Polaroid prints with abstract gradient images, and a single dried gerbera daisy. Editorial photography, shallow depth of field, muted color palette of peach, cream and dusty blue, fine grain, magazine cover composition with negative space on the right.

Generate with this prompt →

flux-schnell/text-to-image, 16:9, $0.005.

"Negative space on the right" reliably leaves room for a headline, and the named palette keeps the colors muted.

Top-down flat lay of a leather-bound notebook, a brass mechanical pencil, a small ceramic espresso cup half full, two cinnamon sticks and a sprig of eucalyptus on a warm oak desk. Soft natural window light from the upper-left, gentle shadows, autumn morning mood. Product photography for stationery e-commerce, square crop, deep contrast, rich brown and olive color palette, no text, no logos.

Generate with this prompt →

flux-schnell/text-to-image, 1:1, $0.005.

Keep "no text, no logos" in product prompts. Without it, Schnell often puts made-up lettering on the notebook cover.

Semi-realistic painted portrait of a young woman with short auburn hair, freckles, gentle smile, looking slightly off-camera. She wears a moss-green knitted turtleneck and small gold hoop earrings. Background is a soft watercolor wash of teal and ochre. Stylized digital illustration with visible brush texture, warm rim light, shallow depth of field, head-and-shoulders crop, profile-picture composition.

Generate with this prompt →

flux-schnell/text-to-image, 1:1, $0.005.

A painted style hides Schnell's weak spots in skin and anatomy better than photoreal faces do.

Where Schnell isn't the right pick: more than two or three words of text in the image, a clean photoreal face for a product page, or an aspect ratio outside the five. For text, Z-Image at $0.008 is the next step up; see the Z-Image API guide.

Draft cheap, render the keeper once

# pip install requests; export HIAPI_KEY=your-key first
import os, time, requests

API = "https://api.hiapi.ai/v1/tasks"
H = {"Authorization": f"Bearer {os.environ['HIAPI_KEY']}"}

def generate(model, **inp):
    r = requests.post(API, json={"model": model, "input": inp}, headers=H, timeout=30).json()
    if r.get("code") != 200:
        raise RuntimeError(r)
    task_id = r["data"]["taskId"]
    while True:
        d = requests.get(f"{API}/{task_id}", headers=H, timeout=30).json()["data"]
        if d["status"] == "success":
            return d["output"][0]["url"]
        if d["status"] == "fail":
            raise RuntimeError(d.get("error"))
        time.sleep(3)

variants = [
    "Ceramic mug on a walnut table, morning window light, no text, no logos",
    "Ceramic mug on white marble, soft overhead light, no text, no logos",
    "Ceramic mug held in two hands, cozy sweater, no text, no logos",
]

# 1. Drafts: 3 × $0.005 = $0.015
drafts = [generate("flux-schnell/text-to-image", prompt=p, aspect_ratio="1:1") for p in variants]

# 2. Look at the drafts, pick one prompt, then render it once on a stronger model
final = generate("gpt-image-2/text-to-image", prompt=variants[0], resolution="1K")  # $0.03

Three drafts and one final cost $0.045. Running all three on the final model would have cost $0.09. The saving grows with the number of variants you try. Every model uses the same POST /v1/tasks call; only model and the fields in input change, and each model page lists its fields.

Billing on HiAPI

• Pay as you go. No subscription and no minimum. You're billed per image, per second of video or per clip, depending on the model.
• One key for every model. Switching from a draft model to a final model doesn't need a second account.
• Failed tasks aren't charged. A request rejected for bad parameters returns a 400 and creates nothing.

FAQ

What's the cheapest image API on HiAPI?

By list price, FLUX.2 Klein 4B: $0.00036 for a 0.25 MP image and $0.00143 at 1 MP. For general use at normal sizes, FLUX.1 Schnell at $0.005 and Z-Image at $0.008 are the usual starting points.

What's the cheapest video API?

Grok Imagine at $0.015 per second at 480p. A 6-second clip is $0.09.

Is a subscription ever cheaper?

Only at a high, steady monthly volume, and most subscriptions are priced for a person using an app, not for generating images inside your own product. For irregular or embedded use, pay-as-you-go fits better.

Do cheap models handle text in images?

Z-Image did well on short English text in our tests. FLUX.1 Schnell doesn't. For Chinese or multi-line layouts, look at Qwen Image.

For video on a budget, see the Hailuo 2.3 API guide and the Grok Imagine video API guide.