GPT Image 2.5
GPT Image 2.5 is OpenAI's model for making and editing images from a prompt and reference images. Flare is the fast choice for most work; Sunburst takes longer but is stronger at edits and at keeping reference subjects.
Variants
Each variant has its own model ID and prices. Pick one, and the prices, playground and code below switch to it.
| Variant | From | View |
|---|---|---|
Flare gpt-image-2.5-flare | $0.001 / image | |
Sunburst gpt-image-2.5-sunburst | $0.001 / image |
GPT Image 2.5 Flare
gpt-image-2.5-flare
Prices by route
| Route | Spec | Price | Official | You save |
|---|---|---|---|---|
Economy Lowest price. Rendered with GPT Image 2.6, GPT Image 2 and 2.5 requests included; 2K and 4K are upscaled from 1K. | Resolution 2K / 4K(upscaled) | $0.001 / image | $0.02676 / image Quality medium | 96% |
| Otherwise | $0.001 / image | $0.01317 / image Quality medium | 92% | |
Express Low price, through ChatGPT. Rendered with GPT Image 2.6, GPT Image 2 and 2.5 requests included; 2K and 4K are upscaled from 1K. No transparent backgrounds or SVG; GIFs always have 9 frames. | Resolution 2K / 4K(upscaled) | $0.002 / image | $0.02676 / image Quality medium | 92% |
| Otherwise | $0.002 / image | $0.01317 / image Quality medium | 84% | |
OpenAI 3 GPT Image on the OpenAI API. | Every request | Text $2.425374 · Image input $3.880598 · Output $14.552239 per 1M tokens ≈ $0.006389 / image at default settings; settled on the tokens used | $0.01317 / image Quality medium | 51% |
Azure 7 GPT Image on Azure OpenAI. | Every request | Text $2.716418 · Image input $4.346269 · Output $16.298508 per 1M tokens ≈ $0.007156 / image at default settings; settled on the tokens used | $0.01317 / image Quality medium | 45% |
Stable Several upstreams tried in turn, cheapest first, so a request rarely fails. | Every request | Text $3.82356 · Image input $6.117696 · Output $22.94136 per 1M tokens ≈ $0.010072 / image at default settings; settled on the tokens used | $0.01317 / image Quality medium | 23% |
Plus Synchronous image generation. GPT Image 2.5 and 2.6 Flare and Sunburst go through Adobe and render 4K natively. No transparent backgrounds or SVG; GIFs always have 9 frames. | Every request | $0.0195 / image | $0.01317 / image Quality medium | — |
Adobe GPT Image through Adobe, rendered natively at 1K, 2K and 4K. No SVG, and no GIFs from reference images. | Every request | $0.0233 / image | $0.01317 / image Quality medium | — |
A task costs the price of the route that serves it. If a route fails, the task moves to the next route you allow and the charge changes to that route's price. Routes priced per token are charged an estimate up front and settled on the tokens actually used. Failed tasks are refunded in full.“Upscaled” means the route renders at a lower resolution and enlarges the image to the size asked for.
Price calculator
| Route | Estimate |
|---|---|
Economy | $0.001 |
Express | $0.002 |
OpenAI 3 ≈ 0 text · 0 image input · 439 output tokens | ≈ $0.006389 |
Azure 7 ≈ 0 text · 0 image input · 439 output tokens | ≈ $0.007156 |
Stable ≈ 0 text · 0 image input · 439 output tokens | ≈ $0.010072 |
Plus | $0.0195 |
Adobe | $0.0233 |
Estimates follow the same rules as billing. A route priced per token charges this estimate when the task is submitted and settles on the tokens the upstream reports when it finishes: the difference is refunded or charged. If your balance can't cover it, the balance goes negative until your next top-up.
Input
PNG, JPEG or WebP, up to 10 MB each. Uploads are kept for 72 hours.
Fill in the required fields to see the price.
Request body
{
"model": "gpt-image-2.5-flare",
"input": {
"resolution": "1K",
"aspect_ratio": "auto",
"quality": "auto"
}
}Output
Run the model to see the result here.
| Field | Description | Allowed values | Default |
|---|---|---|---|
| prompt | What to generate. Required. | — | — |
| negative_prompt | What to avoid. | — | — |
| image_urls | Reference images, or the image to edit: http(s) URLs or base64 data URLs. | — | — |
| resolution | Output size: 1K, 2K or 4K. | 1K, 2K, 4K | 1K |
| aspect_ratio | Width to height, e.g. 1:1, 16:9, 9:16; each model lists its own. auto follows the first reference image: GPT Image draws its exact shape (up to 3:1), other models the nearest ratio offered; 1:1 without one. Very long ratios such as 1:4 or 1:8 are used only when asked, and very tall images may come back as several pictures stacked. | auto, 1:1, 4:3, 3:4, 3:2, 2:3, 16:9, 9:16 | auto |
| count | Images per task, 1–4. Priced per image. | 1, 2, 3, 4 | — |
| quality | auto, low, medium or high. GPT Image only (priced per quality); Nano Banana ignores it. | auto, low, medium, high | auto |
| output_format | png, jpeg or webp; GPT Image 2.6 also gif (animated) and svg (vector, text to image only). Nano Banana ignores it (its files are almost always JPEG); GPT Image on the express and adobe routes returns PNG for png, jpeg and webp whatever is asked. svg is not available on the express, adobe and plus routes, and a GIF from reference images not on adobe: such tasks go to your other routes. Each file's mime_type says what it is; an SVG link downloads when opened. | png, jpeg, webp | — |
| seed | A fixed seed, for repeatable results. Nano Banana ignores it. | — | — |
Submit a task, then poll it until it finishes, or pass callback_url to have the result posted to you.
curl https://api.janvia.io/v1/tasks \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-image-2.5-flare",
"input": {
"prompt": "A red fox in the snow, cinematic light",
"resolution": "1K",
"aspect_ratio": "auto"
}
}'
# Then poll until "status" is "completed" or "failed":
curl https://api.janvia.io/v1/tasks/TASK_ID -H "Authorization: Bearer $API_KEY"