GPT Image 1.5 I2I: AI Image Editor Online
Bottom line
GPT Image 1.5 editing: OpenAI's instruction-following editor, up to 10 photos, 1:1, 2:3 or 3:2, medium or high quality, a flat 10 credits per edit. Open the Image Edit tool, select GPT Image 1.5 I2I, upload your image, and the exact credit cost appears before the edit runs.

GPT Image 1.5 image-to-image is OpenAI's earlier GPT Image editor, on VdoBloom since December 2025. You upload one to ten photos, write the change, and pick medium or high quality. It costs a flat 10 credits per edit, and the quality switch does not change the price, so there is no cost reason to stay on medium.
The spec sheet is deliberately narrow. There are three aspect ratios (square 1:1, portrait 2:3, landscape 3:2, with 3:2 the default), prompts up to 3,000 characters, and no resolution selector. GPT Image 2 and GPT Image 2.5 replaced most of that: they cost less per edit at 1K and 2K, offer 11 to 13 ratios and take 20,000-character prompts. GPT Image 1.5 stays for workflows built around it and for its quality switch.
New accounts get 10 free credits, which covers exactly one edit. A one-time 300-credit pack is $19.50 and never expires (30 edits), or the Lite plan is $15 a month. Below: specs, worked costs, the comparison with GPT Image 2 and 2.5, where it runs, prompting notes and limits.
GPT Image 1.5 editing specs at a glance
The limits below are what VdoBloom's GPT Image 1.5 service enforces. The quality switch is the one control this model has that the newer GPT Image models do not expose.
| Spec | GPT Image 1.5 image-to-image on VdoBloom |
|---|---|
| Maker | OpenAI, run through VdoBloom's KIE integration |
| Mode | Image editing: at least one uploaded photo is required |
| Reference images | 1 to 10 per edit |
| Aspect ratios | 3: 1:1, 2:3, 3:2 (default 3:2) |
| Quality | Medium (default) or high; same price |
| Resolution control | None |
| Prompt length | Up to 3,000 characters |
| Accepted uploads | JPEG, PNG or WebP |
| Price | 10 credits per edit, flat (key gpt-image/1.5-image-to-image) |
| Free credits cover | 1 edit |
| Content filter | STRICT label: OpenAI's provider-side rules |
| On VdoBloom since | 16 December 2025 |
What GPT Image 1.5 costs: worked examples
Every edit is 10 credits whatever the quality, ratio or number of references. The dollar figures use the Lite plan rate of $0.05 a credit; a one-time 300-credit pack is $19.50 and never expires. For comparison, the same edit on GPT Image 2 is 3, 5 or 8 credits depending on resolution.
| Job | GPT Image 1.5 | GPT Image 2 at 1K | GPT Image 2 at 4K |
|---|---|---|---|
| 1 edit | 10 cr ($0.50) | 3 cr ($0.15) | 8 cr ($0.40) |
| 10 edits | 100 cr ($5.00) | 30 cr ($1.50) | 80 cr ($4.00) |
| Medium vs high quality | Both 10 cr | n/a | n/a |
| Edits per 300-credit pack | 30 | 100 | 37 |
| Free signup credits cover | 1 edit | 3 edits | 1 edit |
Read more: GPT Image 1.5 pricing explained · Credit packs and plans
GPT Image 1.5 vs GPT Image 2 vs GPT Image 2.5 Sunburst
All three are OpenAI editors in the same dropdown. GPT Image 1.5 is the flat-price, three-ratio model with a quality switch. GPT Image 2 is resolution-priced (3 / 5 / 8) with 11 ratios and a 20,000-character prompt. GPT Image 2.5 Sunburst keeps that price, raises references to 16 and is tuned for edit precision. VdoBloom does not rank them on output quality; for most new work the newer models cost less, and the honest reason to choose 1.5 is a prompt or look you already rely on.
| Model | Max references | Aspect ratios | Prompt cap | Credits per edit |
|---|---|---|---|---|
| GPT Image 1.5 | 10 | 3 (1:1, 2:3, 3:2) | 3,000 | 10 flat (medium or high) |
| GPT Image 2 | 10 | 11 incl. auto | 20,000 | 3 / 5 / 8 (1K / 2K / 4K) |
| GPT Image 2.5 Sunburst | 16 | 13 incl. auto | 20,000 | 3 / 5 / 8 (1K / 2K / 4K) |
| GPT Image 2.5 Flare | 16 | 13 incl. auto | 20,000 | 3 / 5 / 8 (1K / 2K / 4K) |
Read more: GPT Image 2 editing model page · GPT Image 2.5 Sunburst editing model page · GPT Image 2 text-to-image model page · GPT Image 1.5 vs GPT Image 2 for editing
Medium or high quality?
Both settings cost 10 credits, so the choice is about time and detail, not money. High spends more rendering effort and is the setting to use for a final, for fine texture or for small text. Medium is quicker for checking whether an instruction lands at all. Because the price does not move, a reasonable routine is to test the wording on medium and run the keeper on high. If you need a specific output size rather than a quality level, use GPT Image 2, which offers 1K, 2K and 4K.
Packaging, signage and mock-up edits
Edits that involve words are where an instruction-following editor earns its keep. Typical jobs: changing the flavour name on a can while keeping the brand mark, swapping the price on a shop sign, placing a client's logo on a tote-bag mock-up, or translating a label while keeping the layout. For these, upload the clean product photo first and any logo or artwork as a second image, then write the text exactly as it must appear, in quotes. Tell the model what must stay fixed: the shape of the pack, the existing brand mark, the shadow. Run it on high quality, since small lettering benefits most from the extra rendering effort and the price does not change. If the deliverable needs to be wider than 3:2 or larger than this model produces, take the same prompt to GPT Image 2 at 2K or 4K.
Where to run GPT Image 1.5 on VdoBloom
The model and price are the same everywhere it appears.
- Image Edit: pick GPT Image 1.5 I2I in the edit dropdown, upload 1 to 10 photos, choose 1:1, 2:3 or 3:2 and medium or high.
- Pose Variations: listed among the reference-image models.
Read more: AI Image Editor from a photo · Step-by-step: how to use GPT Image 1.5 · AI pose variations from one photo
Prompting GPT Image 1.5
Write the edit as ordered steps, then a 'keep' list: '1. Change the jacket to dark denim. 2. Keep the logo on the chest intact and legible. 3. Match the lighting from the left. Keep the face, pose and background exactly.' The 3,000-character cap is generous for one edit but tight for long style guides, so cut adjectives before you cut conditions.
Pick the ratio to match the deliverable: 2:3 for a portrait post, 3:2 for a landscape banner, 1:1 for a marketplace tile. For other shapes, generate at the nearest ratio and crop, or switch to GPT Image 2. Quote any text that must appear and spell it exactly. Set quality to high for the final; it costs the same.
When GPT Image 1.5 is still the right pick
For most new edits, GPT Image 2 or 2.5 costs less and offers more shapes. GPT Image 1.5 still earns its place in a few situations.
- You have prompts that were written and tested on GPT Image 1.5 and the results are what you want; changing model means re-testing them.
- You want an explicit medium or high quality setting rather than choosing a pixel size.
- Your outputs are always 1:1, 2:3 or 3:2, so the narrow ratio list costs you nothing.
- The edit has several conditions that must all hold.
Moving a GPT Image 1.5 workflow to GPT Image 2
If you decide to switch, most settings carry over directly. The prompt can stay as it is, since GPT Image 2 accepts far longer prompts. Map the ratio and choose a resolution in place of the quality setting.
| GPT Image 1.5 setting | GPT Image 2 equivalent | Credits on GPT Image 2 |
|---|---|---|
| 1:1, medium | 1:1 at 1K | 3 |
| 1:1, high | 1:1 at 2K (no 4K for squares) | 5 |
| 2:3 or 3:2, medium | Same ratio at 1K | 3 |
| 2:3 or 3:2, high | Same ratio at 2K or 4K | 5 or 8 |
| Up to 10 references | Up to 10 references | No change |
Limits and honest caveats
Three aspect ratios only, no resolution selector and no transparent-background option. Prompts stop at 3,000 characters. At 10 credits it costs more per edit than GPT Image 2 at any resolution (3, 5 or 8), and the 10 signup credits run exactly one edit, so it is not the economical place to experiment. It is the older generation: GPT Image 2 and 2.5 are the actively extended models.
The STRICT label reflects OpenAI's provider rules, so edits involving real people, suggestive content or violence are declined more often than on models labelled HOT, such as Seedream 4.5 Edit. If a job is refused or fails, VdoBloom refunds the credits automatically. It is proprietary, used through VdoBloom's hosted route without a ChatGPT login.
What creators use GPT Image 1.5 I2I for
- Edits with several conditions that all have to survive: change the colour, keep the label text, match the shadow direction.
- Composites that draw on up to 10 photos for one instruction.
- Packaging, signage and mock-up edits where legible text matters more than output size.
- Pose Variations, where it is one of the accepted reference-image models.
- Teams with existing prompts tuned for GPT Image 1.5 who want consistent results without re-testing on a newer model.
How to use GPT Image 1.5 I2I on VdoBloom
- 1Open the Image Edit tool and select GPT Image 1.5 I2I.
- 2Upload the image(s) to edit and describe the change you want.
- 3Review the credit cost and generate. Results appear in your library.
Frequently asked questions
How much does GPT Image 1.5 image-to-image cost on VdoBloom?
A flat 10 credits per edit. Medium or high quality, any of the three ratios, one reference or ten: the price stays 10. At the Lite rate of $0.05 a credit that is $0.50 an edit.
Is GPT Image 1.5 free to try?
The 10 free signup credits cover exactly one GPT Image 1.5 edit. For more test runs at the same budget, GPT Image 2 at 1K costs 3 credits, so the free credits cover three edits there.
Does the medium vs high quality setting change the price?
No. Both cost 10 credits. High spends more rendering effort and suits finals and small text; medium is quicker for drafts.
How many images can I attach to one edit?
Up to 10. The model treats them as joint context for one instruction; give each photo a role in the prompt.
Why does GPT Image 1.5 only offer 1:1, 2:3 and 3:2?
Those are the three ratios the model accepts through VdoBloom's integration (3:2 is the default). For other shapes, crop afterwards or use GPT Image 2 (11 ratios) or GPT Image 2.5 (13).
GPT Image 1.5 vs GPT Image 2: which should I use?
For most new work, GPT Image 2: it costs 3, 5 or 8 credits by resolution against 1.5's flat 10, has 11 ratios and allows 20,000-character prompts. Choose GPT Image 1.5 when you already have prompts tuned for it or want its medium/high quality switch.
Does GPT Image 1.5 support transparent backgrounds?
Not on VdoBloom. Its settings are aspect ratio and quality only. For transparent output, Qwen Image 2.1 offers a transparent setting, and the Die-cut PNG tool uses GPT Image 2.5 Sunburst.
Will GPT Image 1.5 refuse my prompt?
It can. It carries the STRICT label, reflecting OpenAI's provider rules, so requests involving real people, violence or suggestive content are declined more often than on models labelled HOT. Refused or failed jobs are refunded automatically.
What is the maximum prompt length?
3,000 characters; longer prompts are rejected. GPT Image 2 and 2.5 accept up to 20,000.
When did GPT Image 1.5 arrive on VdoBloom?
16 December 2025. GPT Image 2 followed on 21 April 2026 and GPT Image 2.5 on 8 September 2026; all three remain in the edit dropdown.
GPT Image 1.5 I2I guides, prices and comparisons
This page is the GPT Image 1.5 I2I hub. Each guide below goes deeper on one question and links back here.
GPT Image 1.5 pricing
The flat 10-credit price in dollars.
How to use GPT Image 1.5
Upload, instruct, pick a ratio and quality.
GPT Image 1.5 vs GPT Image 2
Flat price against a resolution ladder.
GPT Image 2 editing pricing
3, 5 or 8 credits by resolution.
GPT Image 2.5 guide
Flare and Sunburst, 16 references.
AI pose variations
One photo, new poses, with a reference-image model.
Why ChatGPT image prompts get refused
What the STRICT tier means.
AI Image Editor from a photo
Every edit model in one dropdown.