GPT Image 2. Precise edits, legible text.
OpenAI's high-fidelity model for making and editing images, with the GPT Image 2.5 tiers beside it: Sunburst for precision, Flare for speed. Start from a sentence, or attach a photo and describe the change.
- Text and image input
- Up to 10 reference images
- 8 aspect ratios plus auto
- Quality from low to high, max on 2.5
- Transparent background on 2.5
GPT Image 2 is OpenAI's image model for generation and editing. It takes a text prompt, with reference images when the work has to start from something real, and it follows an edit instruction without redrawing the rest of the frame: ask for the same street at night and the buildings, the subject, and the framing stay where they were. The GPT Image 2.5 series adds two tiers, Sunburst for detailed creative work that needs precision, and Flare for high-volume everyday images that need speed.
On Treza all three sit in the Image Generation node's model picker, with a Quality setting from low to high, extra high and max on the 2.5 tiers, and a transparent background option on 2.5. Draft on Flare at low quality, then re-run the keeper on Sunburst or GPT Image 2 at high. A finished still wires into the First frame port of a video node, and the whole pipeline publishes as one API.
Edits that leave the rest alone
Attach an image and describe the change. The model edits what you asked for and keeps the composition, the subject, and the framing of the original.
Text that reads as written
Signs, labels, and short lines of copy come back spelled the way the prompt spells them, so a storefront or a package mockup does not need a second pass.
Transparent backgrounds
GPT Image 2.5 Sunburst and Flare can return a transparent PNG, a product cutout or an icon ready to place over any layout.
Up to 10 references at once
Wire a product, a person, and a setting into one node and combine them in a single image, or keep a character consistent across a set.
What to make with GPT Image 2
Prompts to try, and what to expect back. Each runs on an Image Generation node with GPT Image 2, GPT Image 2.5 Sunburst, or GPT Image 2.5 Flare picked in the model menu.
“This storefront (attached) at night, rain on the street, keep everything else”
The same scene, relit, with the rest of the frame intact.
“A hand-painted cafe sign on a brick wall that reads OPEN DAILY 7AM”
A photographic still with the lettering spelled as written.
“This sneaker (attached) as a cutout on a transparent background, on Flare”
A PNG ready to place on any layout.
“Four quick concepts for a cooking video thumbnail, Flare at low quality”
A batch of drafts to choose a direction from.
“The chosen concept again on Sunburst at high quality, 16:9”
The keeper, finished with more detail.
“Dress the person in this photo (attached) in this jacket (attached)”
One edit that combines both references.
GPT Image 2 is one node on a canvas
Chain it with language, image, and video models, then publish the whole pipeline as an API. See how a full AI video pipeline fits together.
- Step 01
Draft the brief
A language model node expands a one-line idea into a detailed, on-brand prompt before it ever reaches the model.
- Step 02
Generate with GPT Image 2
Run variants in parallel, compare against other models on the same prompt, and keep every result in run history.
- Step 03
Publish it as an API
The pipeline becomes a versioned endpoint. Your product calls it with the typed /invoke API or any OpenAI SDK.
GPT Image 2 as an API, priced per image
The GPT Image 2 API
Build a pipeline with GPT Image 2 on a node and publish it. It becomes a versioned HTTP endpoint: the typed /invoke route enqueues a run and returns a run id to poll, and an OpenAI-compatible route works with SDKs you already use. Agents connect over MCP the same way. Your prompt engineering ships inside the pipeline, so every caller gets your quality, and swapping the model later does not change the API your product calls.
GPT Image 2 pricing
Prepaid credits, no subscription. Each image is charged at GPT Image 2's own rate from your balance, shown before you run, and only successful generations are charged, so a failed run costs nothing. Credit packs start at $5 and never expire, which suits spiky production schedules. Run history records the cost of every generation, per node, so spend is auditable rather than a surprise.
GPT Image 2, answered
What is the difference between GPT Image 2 and GPT Image 2.5?
GPT Image 2.5 comes as two tiers. Sunburst is the precision tier, suited to detailed creative work, and Flare is the speed tier, suited to high-volume everyday images. Both add extra high and max quality settings and a transparent background option. GPT Image 2 generates and edits at low to high quality. All three take text and image input with the same eight aspect ratios plus auto.
Can GPT Image 2 edit my own photos?
Yes. Drop the photo on the node or wire it into the Reference image port, then describe the change in the prompt. Up to 10 reference images fit on one node, so a product, a person, and a setting can go into a single edit.
Which quality setting should I use?
Start at low for drafts: it is the fastest and the least expensive. Move to medium or high for the final, and to extra high or max on the 2.5 tiers when fine detail has to hold up at a large size. Leave the setting on Model default and OpenAI chooses for you. Each image is charged at the rate for the quality it rendered at.
How much does GPT Image 2 cost on Treza?
Treza uses prepaid credits with no subscription. Each image is charged at GPT Image 2's rate from your balance, and only successful generations are charged. Credit packs start at $5 and never expire.
Can I call GPT Image 2 through an API?
Yes. Build a pipeline that uses GPT Image 2 and publish it as a versioned HTTP endpoint. Call the typed /invoke API from your product, or point any OpenAI SDK at the OpenAI-compatible endpoint. If a better model ships later, swap it on the node without changing the API your product calls.
Your first video is one prompt away.
Generate your first video, image, or draft today. Prepaid credits, no subscription.