Seedance 1.0 Pro Fast (2026): Pricing, Specs and API Code

Seedance 1.0 Pro Fast is the fast-inference version of ByteDance’s Seedance 1.0 Pro video model. You give it one still image and a short prompt describing the motion, and it returns a 5 or 10 second clip at 720p or 1080p. It exists for the situation most teams actually face, where cost per video matters as much as peak quality. Below you will find the exact prices on Unicorn API, how it stacks up against standard Pro and Lite, a complete code example, and the mistakes that quietly burn credits.

What is Seedance 1.0 Pro Fast?

Seedance 1.0 is ByteDance’s first-generation family of AI video models, and it comes in tiers. Seedance 1.0 Lite is the budget tier. Seedance 1.0 Pro is the quality tier, described as producing 1080p video with multi-shot transitions and cinematic motion. Seedance 1.0 Pro Fast sits next to Pro as a variant tuned for faster, cheaper inference. Inference is the work a model does to produce an output once it is trained, so a “fast inference” variant is one built to spend less compute on each video.

On Unicorn API, the model is listed as Seedance 1.0 Pro-Fast-Image to Video with the model id bytedance-v1-pro-fast-image-to-video. It was released on 2025-11-03, a few months after the original Pro and Lite models arrived in August 2025.

Image-to-video only

Image-to-video (often shortened to I2V) means the model takes a still picture as its starting point and animates it, so the subject, lighting and composition of your image carry into the clip. Text-to-video (T2V) starts from words alone. The Pro Fast endpoint is image-to-video: every request needs an image_url. If you only have a text idea, either create a still first or use Seedance 1.0 Pro-Text to Video, which works from a prompt alone.

The inputs it accepts

• prompt: what should move, and how the camera should behave.
• image_url: a public URL for the start frame.
• duration: clip length in seconds. The priced options are 5 and 10.
• resolution: 720p or 1080p in the price list.
• nsfw_checker: toggles the content safety check on the output.

That is a short list, and it is short on purpose. Standard Pro and Lite also accept seed (a number that makes results repeatable), camera_fixed (locks the camera so only the subject moves) and, on Lite, an end_image_url for a target last frame. Pro Fast has none of these, and there is no aspect_ratio field either, so your start image does most of the work of setting the framing. You trade fine control for a lower price.

Seedance 1.0 Pro Fast pricing per clip

Most video models on Unicorn API charge per second of output. Seedance 1.0 Pro Fast is different: it charges a flat price per finished clip, set by the duration and resolution you pick. Prices are in credits. On a monthly plan a credit costs $0.01; top-up credits cost $0.015 each and never expire.

Duration	Resolution	Credits	On a plan ($0.01/credit)	Top-up ($0.015/credit)
5 s	720p	16	$0.16	$0.24
5 s	1080p	36	$0.36	$0.54
10 s	720p	36	$0.36	$0.54
10 s	1080p	72	$0.72	$1.08

Any combination outside these four is billed at 16 credits. Two things stand out in this table. First, 5 seconds at 1080p costs exactly the same as 10 seconds at 720p, so at 36 credits you are choosing between sharpness and length. Second, moving from 720p to 1080p at the same length costs a little over twice as much, which is the main reason to draft at 720p and only render finals at 1080p.

In plan terms, the Starter plan ($25 a month, 2,500 credits) covers 156 clips at 5 seconds and 720p, or 69 clips at 5 seconds and 1080p. The Studio plan ($100 a month, 10,000 credits) covers 138 ten-second 1080p clips. Creator ($50) and Pro ($200) scale the same way; the full list is on the pricing page.

If you want certainty before spending anything, the POST /v1/models/price endpoint returns the exact cost of a request before you run it. The request format is in the API documentation. This is useful in apps where end users pick the length and resolution, because you can show them the cost up front.

How does Seedance 1.0 Pro Fast compare with Pro and Lite?

Here are the same four settings priced across the Seedance 1.0 image-to-video options and one close rival, Grok Imagine Video 1.5. For the per-second models, the rate is multiplied by duration and rounded up to whole credits.

Setting	Seedance 1.0 Pro Fast	Seedance 1.0 Pro I2V	Seedance 1.0 Lite I2V	Grok Imagine Video 1.5
5 s, 720p	16	30	23	23
5 s, 1080p	36	70	50	40
10 s, 720p	36	60	45	45
10 s, 1080p	72	140	100	80

At 720p and 1080p, Seedance 1.0 Pro Fast is the cheapest option in this table at every setting. It costs roughly half of standard Pro, and it also comes in below Lite. That surprises people who assume Lite is always the budget pick within the Seedance family.

When standard Pro is still the right call

Seedance 1.0 Pro-Image to Video costs more, but it gives you controls that Pro Fast does not:

• Repeatability. A seed lets you re-run a prompt and get the same result, which matters when a client approves a draft and asks for a small tweak.
• A locked camera. camera_fixed keeps the frame still, useful for product shots where only the product should move.
• Cheap previews. Standard Pro prices 480p at 2.8 credits per second, so a 5 second preview costs 14 credits.
• A text-to-video sibling in the same tier, plus an aspect_ratio field on that text model.

Pro Fast is a cost and speed variant, so for hero shots where quality is the deciding factor, run the same start image through both models and compare with your own eyes. Do not assume either one wins for your footage.

When Lite wins

Seedance 1.0 Lite Image to Video is the cheapest Seedance option at 480p: 2 credits per second, so 10 credits for a 5 second clip. It also accepts an end_image_url, which lets you set both the first and last frame and have the model fill in the motion between them. If you need text-only input at the budget tier, there is Seedance 1.0 Lite Text to Video.

Other video models worth a look

• Grok Imagine Video 1.5 animates a start frame for 1 to 15 seconds at 480p, 720p or 1080p, with aspect ratio options. It can be busy at peak times. Our guide to 15-second Grok Imagine videos covers it in depth.
• Grok Imagine Video 1.0 handles longer clips, from 6 to 30 seconds, from text or an image.
• Gemini Omni Flash 1.2, released today, goes up to 4K at 22 credits per second and accepts first and last frame URLs. At 1080p it costs 10 credits per second.
• Kling 4.0 Flash Keyframe To Video starts from a first frame at 720p for 17 credits per second.

If you are weighing the newer Seedance generation, see Seedance 2.0 API pricing. For a wider survey, the AI video generator API roundup and the image to video AI tool guide compare more options.

How to call Seedance 1.0 Pro Fast from the API

Unicorn API runs video generation as jobs. You create a job with POST /v1/jobs, poll GET /v1/jobs/{id} until the status is succeeded, then read the video URL from outputs[0].url. The same pattern works for every video model, so switching from Pro Fast to Lite or Grok is a one-line change to the model field.

1. Create an API key on the keys page and export it as UNICORN_API_KEY.
2. Host your start image at a public URL the API can fetch.
3. Send the job request with your prompt, image, duration and resolution.
4. Poll the job until it finishes, then download the video.

Create the job with curl

curl -X POST https://api.unicornapi.net/v1/jobs \
  -H "Authorization: Bearer $UNICORN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "bytedance-v1-pro-fast-image-to-video",
    "input": {
      "prompt": "Slow push-in on the mug, steam curling up from the coffee, soft morning light shifting across the table",
      "image_url": "https://example.com/mug.png",
      "duration": 5,
      "resolution": "720p"
    }
  }'

# Then check on it with the id from the response:
curl https://api.unicornapi.net/v1/jobs/JOB_ID \
  -H "Authorization: Bearer $UNICORN_API_KEY"

A complete Python script

This script creates the job, polls every five seconds, and returns the video URL or raises an error if the job does not succeed.

import os
import time
import requests

API = "https://api.unicornapi.net"
HEADERS = {
    "Authorization": f"Bearer {os.environ['UNICORN_API_KEY']}",
    "Content-Type": "application/json",
}


def generate_clip(image_url, prompt, duration=5, resolution="720p"):
    body = {
        "model": "bytedance-v1-pro-fast-image-to-video",
        "input": {
            "prompt": prompt,
            "image_url": image_url,
            "duration": duration,
            "resolution": resolution,
        },
    }
    r = requests.post(f"{API}/v1/jobs", json=body, headers=HEADERS, timeout=30)
    r.raise_for_status()
    job_id = r.json()["id"]

    while True:
        r = requests.get(f"{API}/v1/jobs/{job_id}", headers=HEADERS, timeout=30)
        r.raise_for_status()
        job = r.json()
        status = job["status"]
        if status == "succeeded":
            return job["outputs"][0]["url"]
        if status in ("failed", "canceled", "cancelled"):
            raise RuntimeError(f"Job {job_id} ended with status {status}: {job}")
        time.sleep(5)


if __name__ == "__main__":
    url = generate_clip(
        image_url="https://example.com/mug.png",
        prompt=(
            "Slow push-in on the mug, steam curling up from the coffee, "
            "soft morning light shifting across the table"
        ),
    )
    print("Video ready:", url)

Two practical notes. Download the output promptly and store it yourself rather than treating the returned URL as permanent storage. And if you work inside Cursor, Claude Code or another coding agent, you can call the same models through MCP, as described on the agents page. If you would rather test prompts before writing code, the video studio runs the same models in the browser.

Writing prompts that work with Seedance 1.0 Pro Fast

With image-to-video, the picture already defines what the scene looks like. The prompt’s job is to describe change: what moves, how fast, and what the camera does. Most weak results come from prompts that re-describe the image instead.

• Lead with motion. “Steam rises, light shifts slowly from left to right” beats “a white mug on a wooden table”.
• Pick one camera move. A push-in, a slow pan or a static shot. Stacking “orbit, zoom, then tilt” in 5 seconds tends to produce muddled movement.
• Match the motion to the length. A 5 second clip suits one action. Save multi-step actions for 10 seconds.
• Use a clean, sharp start image. Blur, compression artifacts and cluttered backgrounds get animated too.
• Frame the image for its destination. Since Pro Fast has no aspect ratio field, start from an image already shaped for where the video will be shown, such as vertical for short-form social feeds.

A worked example: a product clip for a store page

Say you sell ceramic mugs and want a 10 second 1080p loop for each product page. Here is a budget-aware way to get there with Seedance 1.0 Pro Fast:

1. Draft. Write three prompt variations and run each twice at 5 seconds and 720p. That is 6 clips at 16 credits, so 96 credits.
2. Extend. Take the two best prompts and run them at 10 seconds and 720p to check the motion holds up over the longer length: 2 clips at 36 credits, so 72 credits.
3. Finalise. Render the winner at 10 seconds and 1080p: 72 credits.

Total: 240 credits, or $2.40 on a monthly plan. Had you rendered all nine of those clips at 10 seconds and 1080p, you would have spent 648 credits for the same final result.

Is Seedance 1.0 free, unlimited or open source?

These are among the most common Seedance 1.0 searches, so here are straight answers.

Free and unlimited: no. Every clip consumes real GPU time, and someone pays for it. Sites that promise free unlimited Seedance generation usually cap daily use, add watermarks, lower the resolution or put free users at the back of the queue. On Unicorn API every clip has a known price, and the cheapest way to try Seedance 1.0 Pro Fast is a 5 second 720p draft at 16 credits ($0.16 on a plan).

Open source and downloads: Seedance 1.0 is offered as a hosted model that you reach through an API, not as weights you download and run on your own GPU. That is why searches for a Seedance download or a Seedance ComfyUI workflow usually lead to integrations that call a hosted API in the background. On Unicorn API there is nothing to install: you send a request and receive a video URL.

Common mistakes with Seedance 1.0 Pro Fast

1. Sending a text-only request. This model needs an image_url. For prompt-only video, use a text-to-video model instead.
2. Rendering every draft at 1080p. 1080p costs over twice as much as 720p at the same length. Draft low, finish high.
3. Expecting seed or camera_fixed to work. They are not inputs for Pro Fast. If you need repeatable output or a locked camera, use standard Pro.
4. Using a private or expiring image URL. Signed URLs that expire in seconds, or images behind a login, cannot be fetched. Use a public URL that stays valid for the life of the job.
5. Starting from the wrong shape. A landscape image will not become a good vertical video. Crop the source first.
6. Re-describing the image in the prompt. Spend your words on motion and camera, not on what is already visible.
7. Comparing prices the wrong way. Pro Fast charges per clip; most rivals charge per second. Convert everything to a per-clip number at the length you actually need, as in the comparison table above, before deciding.
8. Not handling failed jobs. Check for a non-success status in your polling loop so a failed job does not leave your script waiting forever.

For most product clips, social posts and prototypes where you start from a good image, Seedance 1.0 Pro Fast is the cheapest way to get Seedance-quality motion at 720p or 1080p. Reach for standard Pro when you need control, Lite when you need 480p or an end frame, and check the exact price before each batch. You can browse every option in the model catalog.