GPT Image 2 Image-to-Image Pricing: 3, 5 or 8 Credits Per Edit

GPT Image 2 image-to-image is priced by output resolution on VdoBloom: 3 credits at 1K, 5 credits at 2K and 8 credits at 4K — $0.15, $0.25 and $0.40 at the Lite rate where $15 buys 300 credits and one credit is $0.05. The charge is per edit request, not per input image, so the same 3 credits cover a one-photo retouch and a ten-photo composite at 1K.

The three price tiers, and the rules that decide which one you land on

Unlike its flat-priced predecessor, GPT Image 2 gives you a resolution selector, and that selector is the only thing that changes the price. Quality settings, prompt length and input count do not. But two provider constraints quietly control which tier you can actually reach, and it is worth knowing them before you plan a budget:

• Auto aspect ratio forces 1K. If you leave the aspect ratio on auto — which is the default — the request is generated at 1K and billed at 1K, even if you selected 2K or 4K. Pick an explicit ratio to unlock the higher tiers.
• 1:1 square cannot produce 4K. The request is rejected before any credits are deducted, with a message telling you to use 2K or a non-square ratio.
• No resolution supplied means 1K. A request that omits resolution defaults to the 1K variant for billing, which matches what the provider actually renders.

Full pricing table

Resolution tier	Credits per edit	Cost at $0.05/credit	Aspect ratios available	What you get
1K (default)	3	$0.15	All 11, including auto	Web-sized output, cheapest way to iterate
2K	5	$0.25	All 10 explicit ratios (not auto)	Print-usable detail, safe middle tier
4K	8	$0.40	All explicit ratios except 1:1	Maximum detail for large crops and print

The explicit ratio list is generous: 1:1, 5:4, 4:5, 4:3, 3:4, 3:2, 2:3, 16:9, 9:16 and 21:9, plus auto. So you can take a 9:16 vertical to 4K for 8 credits, but a 1:1 square tops out at 2K for 5 credits.

What each budget buys

Credit source	Credits	Edits at 1K (3cr)	Edits at 2K (5cr)	Edits at 4K (8cr)
Free signup balance, no card	10	3	2	1
One-time pack from $2.49	75	25	15	9
Lite, $15/month	300	100	60	37
Lite billed annually, $10/month	300	100	60	37
Pro billed annually, $49/month	2,000	666	400	250

Annual billing does not change the credit cost, it changes what a credit costs you. On annual Lite a credit is about $0.033, so a 1K edit is roughly $0.10 instead of $0.15; on the annual Pro 2,000-credit tier a credit is closer to $0.025, putting a 4K edit near $0.20. One-time packs never expire, which suits bursty work. The live tier list is on the pricing page.

How that compares with the other editors on VdoBloom

At 3 credits for a 1K edit, GPT Image 2 image-to-image is one of the cheapest per-edit image editors on the platform, and its 8-credit 4K tier still undercuts several models’ base price:

Editor	Credits per edit	Cost at $0.05/credit	Best for
GPT Image 2 I2I, 1K	3	$0.15	High-volume iteration on instruction-heavy edits
GPT Image 2 I2I, 4K	8	$0.40	Final-pass detail at a wide or vertical ratio
Seedream 5 Lite I2I / Qwen2 image edit	6	$0.30	Looser content filtering than the GPT family
Seedream 4.5 edit	7	$0.35	Photographic restyles and swimwear or fitness work
GPT Image 1.5 I2I	10 flat	$0.50	Long conditional instructions, flat price regardless of settings
Flux 2 Pro I2I	12	$0.60	Edits that must preserve original source framing
Nano Banana Pro	18 (24 at 4K)	$0.90–$1.20	Premium detail work where price is secondary

The honest comparison against GPT Image 1.5 image-to-image is that GPT Image 2 is cheaper at every tier and carries a far longer prompt limit — 20,000 characters against 3,000 — so for most instruction-driven editing it is the better economics. The reason to keep 1.5 in your rotation is its flat price: if you always work at high quality in 3:2, 10 credits with no tier rules to track is simpler than 8 credits with two constraints attached. Note that both carry the platform’s strictest content label, since OpenAI’s provider filters are conservative; if your work is swimwear, dance or fitness content, the flexible-tier Seedream and Qwen editors are the ones that will actually run it.

What you are not billed for

Input images are free. The model accepts up to 10 per job and VdoBloom charges the request, so a composite pulling a subject, a background and a texture from three separate photos costs the same as a single-image tweak at that resolution. Failed jobs are refunded automatically: a provider error, an empty result, or a moderation block all return the credits to your balance with a message explaining which happened. There is no charge for prompt length, no storage fee, and paid plans download watermark-free with commercial rights.

A worked example

You are preparing 60 lifestyle images for a campaign. The workflow is two passes: a cheap exploration pass to find the right edit, then a final render. Run all 60 explorations at 1K on auto aspect for 3 credits each — 180 credits, about $9 at the Lite rate — then take the 20 keepers to 4K at 16:9 for 8 credits each, another 160 credits. Total 340 credits, which is just over a monthly Lite allowance and roughly 17 percent of the annual Pro 2,000-credit tier. Doing all 60 at 4K instead would cost 480 credits, so the two-pass habit saves 140 credits on a job this size.

Start in the image editor, generate a base image first in image generation if you do not have a source photo, or read the full spec sheet on the GPT Image 2 I2I model page. If your source photo shows a real person, you need that person’s consent before editing or publishing the result.

Frequently asked questions

How much does a GPT Image 2 image-to-image edit cost?

3 credits at 1K, 5 at 2K and 8 at 4K — $0.15, $0.25 and $0.40 at the Lite rate of $0.05 per credit, and roughly $0.075, $0.125 and $0.20 on the annual Pro tier.

Why was my edit billed at 1K when I selected 4K?

Because the aspect ratio was left on auto. The provider forces auto requests to 1K, so VdoBloom bills the 1K tier to match what was actually rendered. Choose an explicit aspect ratio to reach 2K or 4K.

Can I get a 4K square image?

No. 1:1 cannot be generated at 4K. The request is rejected before credits are taken — use 2K for a square, or switch to a non-square ratio such as 4:3 or 16:9 for 4K.

Does the number of input images change the price?

No. Up to 10 input images are included at every resolution tier, because billing is per edit request rather than per image.

Is GPT Image 2 cheaper than GPT Image 1.5?

Yes, at every tier. 1.5 is a flat 10 credits per edit; GPT Image 2 ranges from 3 to 8 credits, and its 1K tier is less than a third of 1.5’s price.

What happens to my credits if the edit is blocked by moderation?

They are refunded automatically. Both GPT Image models sit in the strictest content tier, so refusals happen more often than on the flexible-tier editors, but a refusal never costs credits.