Seedream 4.0 is the entry tier of ByteDance's Seedream family. It is the one that takes a resolution rather than a quality switch, and the one cheap enough to run several times while you are still deciding what you want.
The only Seedream tier here that exposes a resolution ladder directly.
Square through 21:9 landscape, chosen by name rather than by ratio.
Examples made on Kavel with this model.
Generated with Seedream 4.0 on this site on 2026-08-04, unretouched. Both are the kind of subject this tier handles well and the tiers above it are wasted on: a clean subject, a single light source, and nothing in frame that has to be read.
Seedream 4.0 is the oldest tier of ByteDance's Seedream family still worth offering, sitting below 4.5, 5.0 Lite and 5.0 Pro. It is also the odd one out in how you talk to it: where the later tiers take an aspect ratio and a quality setting, 4.0 takes a named framing preset — square, portrait_16_9, landscape_21_9 and so on — plus a resolution of 1K, 2K or 4K.
Only the prompt is actually required. What it is for is the part of the process before you know what you want. Most of the images anyone generates are thrown away, and they are thrown away for reasons that have nothing to do with model tier: the composition was wrong, the idea did not work, the subject was in the wrong place.
Finding that out at eight credits instead of twenty-one is the whole argument for this tier. Where it visibly gives up is text inside the image — 4.5 is the tier that fixed that — and faces under difficult light, which is what 5.0 Pro is for.
Creative engine
One flat rate, and the resolution you pick does not change it.
Below 4.5, 5.0 Lite and 5.0 Pro, and the cheapest way to run the same prompt.
square, portrait_4_3, landscape_16_9 and six more, plus a 1K/2K/4K resolution.
Everything else has a default, which makes it the quickest tier to throw an idea at.
Pick 4.0 when you are still deciding. That covers the first several runs of almost every image anyone makes here: testing whether the composition works, whether the subject reads at a glance, whether the idea was any good. The prompt you arrive at transfers to 4.5 and 5.0 Pro unchanged, so nothing spent down here is wasted. Move up the moment either of the two known weaknesses matters — writing that has to be legible inside the frame, which is 4.5, or a human face doing something difficult, which is 5.0 Pro. And stay here permanently for decorative work: backgrounds, textures, and anything that will sit behind text at low opacity.
Roughly a fifth less than 4.5 and a third less than 5.0 Pro at its basic setting.
The resolution is a request field here rather than something implied by a quality switch.
Whatever wording works here works on 4.5 and 5.0 Pro without rewriting.
What the entry tier accepts, and where the tiers above it pick up.
One prompt in, one image out, with no source material required.
square, square_hd, portrait_4_3, portrait_3_2, portrait_16_9, and four landscape equivalents up to 21:9.
A real resolution field, which the 4.5 and 5.0 tiers replaced with a quality switch.
Reuse a seed to get closer to a previous run rather than starting over each time.
Taken from the published provider schema for this model, which is the API it runs on here.
Four steps, and the last one is what this tier is actually for.
Choose the shape by name before writing — it changes what belongs in the frame.
What the thing is, and where the light comes from. Both do more than any adjective.
Resolution does not change the composition, and 4K is only worth generating once you like it.
At eight credits, three attempts still cost less than one image on the top tier.
Jobs where run count matters more than the last increment of fidelity.
The first several attempts at any image, before the prompt has settled.
Images that will sit behind type at low opacity and never be inspected.
One object, one light source, no writing on the label.
Anywhere you need many images rather than one good one.
One credit pool covers Nano Banana images and Seedance and Veo video. The cost shows before every run, and failed jobs are not charged. Use a subscription for ongoing work, or a one-time pack when you just need to top up.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
One-time top-ups — buy extra credits any time you run low.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
What's included
Secure checkout by Stripe. Card details never touch our servers.
Charges appear as “KAVEL AI” on your card statement. You can cancel any time from Settings → Billing; cancellation takes effect at the end of the period you already paid for.
Powered by
Use one balance across every supported image and video model — ten of them today, listed below — and check the credit cost before each request.
Video models
Image models
Model credit guide
You're only charged for successful generations. The exact estimate in the generator varies by model, length, resolution, audio, and number of images.
| Type | Model | Credit cost |
|---|---|---|
| Video | Seedance 2.0 | 5s 720p image-to-video ≈ 188 credits; text-to-video ≈ 308 credits. Scales with resolution and length. |
| Video | Seedance 2 Fast | Faster and lower cost. 5s 720p text-to-video ≈ 248 credits. |
| Video | Seedance 2 Mini | The cheapest Seedance tier. 5s 720p ≈ 154 credits, 5s 480p ≈ 72 credits. |
| Video | Seedance 1.5 Pro | Audio doubles the rate. 5s 720p ≈ 27 credits silent, ≈ 53 with audio; 1080p ≈ 57 and ≈ 113. |
| Video | Veo 3.1 | Billed per video, not per second (Lite tier). About 45 credits at 720p and 53 at 1080p. |
| Video | Kling 3.0 | Audio raises the rate. 5s 720p ≈ 105 credits silent, ≈ 150 with audio; 1080p ≈ 135 and ≈ 203. |
| Video | MiniMax H3 | Fixed 2K, no resolution ladder. Priced per second — a 5s clip ≈ 158 credits. |
| Image | Nano Banana 2 | Generate or edit from text and images. About 8 credits per 1K image, 12 at 2K, 18 at 4K. |
| Image | Nano Banana Pro | Consistent run times across generations. About 12 credits per 1K or 2K image, 21 at 4K. |
| Image | Nano Banana 2 Lite | Faster, simpler variant. A flat 8 credits per image at every resolution. Start here to test. |
| Image | GPT Image 2 | The lowest-cost image model here. About 3 credits per 1K image, 6 at 2K, 12 at 4K. |
| Image | Seedream 5.0 Lite | Flat pricing — about 8 credits per image whether you generate or edit. |
Common questions about the entry tier of the Seedream family.
4.0 takes a named framing preset and a 1K/2K/4K resolution; 4.5 takes an aspect ratio and a basic/high quality switch. The practical difference is text — 4.5 renders writing inside an image far more reliably — and 4.5 costs a little more.
About 8 credits, and the resolution you pick does not change it. Seedream 4.5 is about 10 and Seedream 5.0 Pro about 11 at its basic setting.
Only once you like the image. Resolution does not change composition or subject, so generating your test runs at 1K and the keeper at 4K loses nothing.
That is this tier's known weakness. Legible text inside an image is what Seedream 4.5 improved; if the words matter, generate there instead.
Reusing a seed keeps a new run close to a previous one, so you can change one part of the prompt without the whole image rearranging itself.
Credits are returned automatically, unless the request broke the content policy.
Pick a framing preset, describe the subject and the light, and run it a few times.
The cheapest tier in the family, which is what makes it the one to iterate on.
