> ## Documentation Index
> Fetch the complete documentation index at: https://docs.toapis.com/llms.txt
> Use this file to discover all available pages before exploring further.
# GPT-Image-2.5 Official Image Generation
> Async Sunburst and Flare image tasks through the official channel, with R2 delivery and token pricing at 80% of official rates
Access both GPT-Image-2.5 models through the official Azure channel. Submission returns an asynchronous task ID immediately. Images are generated in the background and stored in R2 before delivery.
| Model | model |
| ----------------- | --------------------------------- |
| Sunburst Official | `gpt-image-2.5-sunburst-official` |
| Flare Official | `gpt-image-2.5-flare-official` |
Use the full model name from this table. `gpt-image-2.5-official` is a series label, not a valid request model.
For mainland China, replace `https://api.toapis.com` with `https://api.toapis.cn`. Create an API Key in the [console](https://toapis.com/console/token) and set the `TOAPIS_API_KEY` environment variable.
## Submit and Query
Submit one Flare image. Change model to call Sunburst:
```bash theme={null}
curl --fail-with-body --request POST \
--url https://api.toapis.com/v1/images/generations \
--header "Authorization: Bearer $TOAPIS_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
"model": "gpt-image-2.5-flare-official",
"prompt": "A small blue circle on a plain white background",
"quality": "low",
"size": "1024x1024",
"n": 1
}'
```
```json theme={null}
{
"id": "tsk_img_example",
"object": "generation.task",
"model": "gpt-image-2.5-flare-official",
"status": "pending",
"progress": 0,
"created_at": 1789099098,
"metadata": {}
}
```
Save the `id` from the submission response and replace `TASK_ID` below:
```bash theme={null}
curl --fail-with-body \
--url https://api.toapis.com/v1/images/generations/TASK_ID \
--header "Authorization: Bearer $TOAPIS_API_KEY"
```
Submission returns `pending`. Polling may return `queued` or `in_progress`, followed by `completed` or `failed`. On `completed`, read the R2 image URLs from `result.data[].url`. On `failed`, read `error`. Poll the same task ID every few seconds. High-quality generation can take several minutes; keep polling instead of submitting duplicate requests. See the [image task status API](../../tasks/image-status) for the complete response fields.
## Request Parameters
Use a ToAPIs API Key as `Bearer YOUR_API_KEY`. Azure credentials are not required.
`gpt-image-2.5-sunburst-official` / `gpt-image-2.5-flare-official`.
Describe the image. For reference-image requests, describe what to preserve and change.
Supports `low`, `medium`, `high`, `xhigh`, and `max`. Quality and dimensions affect actual token usage and processing time.
Use pixel dimensions such as `1024x1024`, `1536x1024`, or `1024x1536`. Custom dimensions must pass upstream validation. These examples do not require an additional resolution field.
Optional. Set `transparent` to request a transparent background. Omit it to use the upstream default. The default PNG output preserves transparency.
The official channel currently generates one image per request. Use `1`.
## Reference Image URLs
For image-to-image requests, add `image_urls` to the same JSON generation request. URLs must be accessible to the server. You can obtain URLs through the [image upload API](../../uploads/images). Reference images incur image input token charges. Submission and polling use the same asynchronous flow.
```bash theme={null}
curl --fail-with-body --request POST \
--url https://api.toapis.com/v1/images/generations \
--header "Authorization: Bearer $TOAPIS_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
"model": "gpt-image-2.5-sunburst-official",
"prompt": "Keep the subject and replace the background with a snowy forest",
"image_urls": ["https://example.com/reference.png"],
"quality": "high",
"size": "1024x1024",
"n": 1
}'
```
## Token Pricing
Standard prices verified on 2026-09-11. Both official models are billed at 80% of official token rates, a 20% discount:
| Type | USD / 1,000,000 tokens |
| ------------------ | ---------------------: |
| Text input | 4.00 |
| Cached text input | 1.00 |
| Image input | 6.40 |
| Cached image input | 1.60 |
| Image output | 24.00 |
Quota is reserved at submission and reconciled against actual upstream text and image token usage on success. All five quality levels use these unit prices; there is no fixed per-image price for each quality. Sufficient account balance and API Key quota are required before submission.
Cost formula, in USD:
```text theme={null}
USD = (
uncached_text_input_tokens * 4
+ cached_text_input_tokens * 1
+ uncached_image_input_tokens * 6.4
+ cached_image_input_tokens * 1.6
+ image_output_tokens * 24
) / 1,000,000
```
For example, 17 uncached text input tokens and 196 image output tokens cost `(17 * 4 + 196 * 24) / 1,000,000 = $0.004772`. This is not a fixed price per image.
Account-specific pricing or discounts still apply. Verify the final charge in usage logs. Current prices follow the [pricing page](https://toapis.com/pricing) and your account configuration.
## Related Models
The [standard tier](../gpt-image-2.5/generation) and [VIP tier](../gpt-image-2.5-vip/generation) have their own full model names. Use the `-official` suffix for the channel documented here.