GPT Image 2.5 Flare
gpt-image-2.5-flareBuilt for efficient image creation, balancing quality and speed for creative iteration, marketing assets and product concepts.
Review the cost estimate before you run a test.
Illustration- Modality
- Image
- Image sizes
- Automatic
- List price (USD)
- Input 5.000000 · Output 30.000000 / million tokensCache read 2.000000 / million tokens
Give your ideas a visual form.
Describe the subject, style and composition, then create an image with the available settings.

Model specifications and parameters
Fields are shown for this model's connected endpoint, not one generic form for every model.
| Parameter / field | Current range | Description |
|---|---|---|
| size | auto | Resolution presets come from this model's catalogue configuration. |
| Aspect ratio | Describe in prompt | A fixed ratio maps to valid dimensions; smart mode lets the model decide. |
| Custom dimensions | Not enabled here | Use one of this model’s size presets. Arbitrary pixel dimensions are not enabled. |
Model specifications and parameters
Fields are shown for this model's connected endpoint, not one generic form for every model.
| Parameter / field | Current range | Description |
|---|---|---|
| size | auto | Resolution presets come from this model's catalogue configuration. |
| Aspect ratio | Describe in prompt | A fixed ratio maps to valid dimensions; smart mode lets the model decide. |
| Custom dimensions | Not enabled here | Use one of this model’s size presets. Arbitrary pixel dimensions are not enabled. |
| n | 1–1 | Images allowed in one web request; unverified image-set modes remain unavailable. |
| sequential_image_generation | Not enabled here | Available only for image-set models; n bounds the maximum billable output count. |
| image | Optional | An uploaded library image may be used as a reference. |
| response_format | b64_json | Synchronous requests return base64 images. Asynchronous image tasks provide download URLs. |
Provider limits and LINK42 web-test limits may differ. LINK42's current quote and actual settlement govern price; external docs are provided to verify model capabilities.
Test a result before integrating
Adjust parameters, review the estimate and validate your idea with an actual output.
Capabilities & use cases
Create and edit
Use /v1/images/generations for text-to-image or reference editing. Requests with image inputs automatically use the editing transport; /v1/images/edits also accepts JSON. Generate one image with size=auto. References can be owned asset:// IDs, image URLs or Base64 data.
Frequently asked questions
Are playground tests billed?
Yes. Tests use the signed-in account's balance. The estimated hold is shown before submission, and the final amount is calculated from actual usage and effective pricing.
What happens if a test fails or stops?
Requests rejected before generation have no model usage. Failures, cancellations and interrupted streams are settled or released by the shared ledger. Check the run record for any billable usage already produced.
Where can I see results and charges?
Recent results and actual charges are available in the test history. Open the usage record using its request ID. Download large results promptly.
Pricing
These are standard model prices. Signed-in estimates account for your billing identity and parameters; final charges use actual usage.
- Input price (USD)
- 5.000000/ million tokens
- Output price (USD)
- 30.000000/ million tokens
- Cache reads (USD)
- 2.000000/ million tokens
API
After validating the result, use the same model identifier in your application. Keep API keys on your server.
curl https://www.link42.ai/v1/images/generations -H "Authorization: Bearer $LINK42_API_KEY" -H "Content-Type: application/json" -d '{"model":"gpt-image-2.5-flare","prompt":"…","response_format":"b64_json"}'Ready to build with it?
Explore the API reference and code examples to bring this model into your app.