The image model behind ChatGPT's picture generation, available here on its own. Same model, no subscription, and a per-image credit cost you can see before each run — which is the part a chat window does not give you.
Pay per image in credits instead of a monthly plan you may not use.
3 at 1K, 6 at 2K, 12 at 4K — the cheapest image model on the site.
Examples made on Kavel with this model.
AI-generated reference frames chosen around this model's known strength — pictures whose content is partly typographic. Information cards, annotated journals and travel posters all fail immediately if the lettering does not hold. They illustrate the territory rather than being outputs captured from this specific model.
When ChatGPT produces a picture, it is calling OpenAI's image model — the same model available here as GPT Image 2, on its own and without the chat wrapper. Whether that is better than using it inside ChatGPT depends on what gets in your way. The chat window is comfortable and conversational, and it hides two things: what each image costs you, and the controls.
Here the cost of a run shows before you commit to it, resolution is a setting rather than something inferred from your phrasing, and the output arrives as a file rather than as a message you have to save out of a conversation. What you give up is the conversation itself — there is no back-and-forth refining across turns, so the prompt has to carry the intent on its own.
The model's particular strength is text inside the picture: labels on a diagram, headings on a poster, copy on packaging. That is the job it is most often picked for.
Creative engine
Resolution is the only lever on price here, and the range is narrow enough that 2K is an easy default.
OpenAI's image model without a ChatGPT subscription attached to it.
The credit estimate appears before you generate rather than being absorbed into a monthly plan.
In-image text is what this model is most often reached for.
Use it here when you want the controls and the cost visible: a fixed resolution rather than one inferred from your wording, a file rather than a message, and a number in front of you before each run. Use it here too when you only need images occasionally, because a per-image cost beats a monthly subscription that sits idle. Stay in ChatGPT when the conversation is the point — when you expect to refine an image over several turns, or when you want to talk through the idea before committing to a prompt. And reach for a different model entirely when what you need is a multi-photo composite: that is the Gemini line's territory, not this one's.
Resolution and aspect ratio are settings here, not things you have to talk the model into.
Per-image credits suit a few images a month better than a subscription does.
For combining several source photos into one image, the Gemini Flash tier is the right tool.
What the model covers here, stated as what it does rather than what a chat interface makes it feel like.
Generate a picture from a written description with no source material.
Supply a photo and describe the change you want made to it.
1K, 2K or 4K, set explicitly rather than inferred from the prompt.
Output shaped for the place the image is going.
The generation parameters for this model as it runs here.
Four steps, and no conversation to manage.
Generate from a description, or start from a picture you upload.
There is no follow-up turn here, so put the layout, style and any exact wording in from the start.
Choose 1K, 2K or 4K and the aspect ratio, rather than hinting at them in words.
Confirm the estimate, run it, and download the file.
The jobs that play to a model known for handling type and layout.
Figures, headings and blocks arranged into something legible in one pass.
A title that sits inside the composition rather than being added afterwards.
Copy rendered on the pack, which is the part cheaper models tend to garble.
At 3 credits a 1K image this is the cheapest model here, which makes iteration cheap too.
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 using OpenAI's image model on Kavel.
It is OpenAI's image model, GPT Image 2 — the same model ChatGPT calls when it makes a picture. What differs is the wrapper around it: controls and a visible per-run cost instead of a conversation.
No. It is billed per image in credits here, so there is no monthly plan to hold.
3 credits at 1K, 6 at 2K and 12 at 4K, which makes it the cheapest image model on the site. The estimate shows before you generate.
Yes. Supply the photo and describe the change. Name what should stay the same as well, or the edit can drift.
Images with type in them — infographics, posters, packaging, anything labelled. Quote the exact wording in your prompt rather than describing it.
For multi-photo composites, use Gemini Flash Image (Nano Banana 2), which accepts up to 14 reference photos in a single run.
Open the generator, write your prompt, pick a resolution, and generate. No subscription.
The model's known strength: labels, headings and layout that read as designed.
