GPT Image 2.5 Sunburst
OpenAI GPT Image 2.5 Sunburst, the precision tier of GPT Image 2.5. Slower than Flare, but holds faces, fine geometry and small text. Same price as Flare at every quality and size, so draft on Flare at medium and ship here at high. xhigh and max cost more credits than high (max is 2 to 4x); use them for final hero frames. Sizes up to 2560x1440 are reliable; larger is experimental, so upscale the keeper with Crystal Upscaler (creativity 0) instead.
Try GPT Image 2.5 Sunburst in the playground
GPT Image 2.5 Sunburst
Node type: generate:openai_gpt_image_2_5_sunburst_text_to_image
Category: generate
Description
GPT Image 2.5 Sunburst is the precision tier of OpenAI's GPT Image 2.5: write a Prompt and run the node for 1 to 4 images that hold faces, fine geometry and small text, at a slower pace than Flare. Its price matches the GPT Image 2.5 Flare node at every quality and size, so draft on Flare at medium and make the final here at high. It costs about 17 credits per image; xhigh and max cost more (max is 2 to 4x), so keep them for hero frames. Sizes up to 2560x1440 are reliable, so for more pixels, upscale the keeper with the Crystal Upscaler node at creativity 0. Wiring an image in routes the run to Sunburst Edit.
Pricing
- Cost: ~17 credits per images
Variant family
- Family:
gpt-image-2.5-sunburst - Is primary: yes
- All variants:
generate:openai_gpt_image_2_5_sunburst_text_to_image,edit:openai_gpt_image_2_5_sunburst_edit
Canvas ports
These appear as port handles on the left side of the node.
| ID | Label | Details |
|---|---|---|
prompt |
Prompt | TEXT (required) · multiline |
image1 |
Image 1 | IMAGE |
Sidebar config
These render as form fields in the right-side config panel when the node is selected.
| ID | Label | Details |
|---|---|---|
quality |
Quality | TEXT · options: low, medium, high, xhigh, max |
image_size |
Image Size | IMAGE_SIZE · options: custom, square_hd, portrait_4_3, landscape_4_3, auto — Up to 2560x1440 is reliable. Larger custom sizes are experimental and can come out over-sharpened; upscale instead when a crop needs more pixels. |
num_images |
Number of Images | NUMBER · 1…4 |
output_format |
Output Format | TEXT · options: jpeg, png, webp |
background |
Background | TEXT · options: auto, transparent, opaque |
output_compression |
Output Compression | NUMBER · 0…100 |
sync_mode |
Sync Mode | BOOLEAN |
Outputs
| ID | Label | Type |
|---|---|---|
images |
Generated Images | IMAGE |
Realistic avatar stills
Building a talking-head avatar? Make a Realistic AI Avatar has the tested recipe: one pass at custom 1152x2048, no upscale on a still that feeds Seedance, and no grain wording in the prompt. Wiring any image here routes to the edit node below, so keep identity refs off image1.
Auto-generated from the Wireflow node registry.