GPT Image 2.5 on SceneMixer: Sunburst and Flare for AI drama sheets and covers
Choosing between the two: Sunburst vs Flare on SceneMixer. Against Google’s top tier: GPT Image 2.5 vs Nano Banana Pro.
What GPT Image 2.5 is
OpenAI launched ChatGPT Images 2.5 on September 8, 2026, and with it two API models: Flare and Sunburst. OpenAI describes Flare as the general-purpose model, faster than the previous generation, and Sunburst as the slower one built for precise work, keeping earlier edits and reference subjects intact across turns. Those are OpenAI’s own claims; the sources are at the end of this page.
On SceneMixer you do not need an OpenAI account, an API key or a ChatGPT plan. The two models are tiers inside the site and are billed in SceneMixer credits, like every other tier.
What it draws on SceneMixer
An AI drama series stays consistent because every shot is generated against the same reference images. SceneMixer reads your script, lists the cast, locations and props, and draws a sheet for each: a full-length casting photo on white for a character, an empty room with its light source readable for a location, the object alone on white for a prop. A second look — the wedding gown, the cast on a broken arm — becomes a state image drawn from the main sheet. The image tier you pick decides how those are drawn.
The same tier also paints the promo cover: three poster candidates with your title painted into the picture, built from your leads’ sheets. See make a promo cover poster for your AI drama.
It does not make the video. Shots are rendered on the video tiers; GPT Image 2.5 draws the pictures they are matched to.
Where to switch it on
Settings → Model Preferences → Image Model, then pick GPT Image 2.5 Sunburst. From then on every new image is drawn on it: the character, location and prop sheets in step 2, regenerations, state images, and the promo cover posters. Images already drawn keep their pixels; switching never redraws anything. You can switch per group — leads on one tier, background places on another — and switch back.
- Open Settings from the account menu.
- Go to Model Preferences → Image Model.
- Choose GPT Image 2.5 Sunburst or GPT Image 2.5 Flare. The price per image is shown next to each tier.
- Back in your project, generate or regenerate the sheets in step 2.
Sunburst or Flare
On this site the two cost the same, 25 credits per image, and draw at the same size, so the choice is about what each model is built for.
- Flare for drafting and for volume: the whole supporting cast, background locations, props. It is the faster of the two.
- Sunburst where fidelity to a reference matters: state images drawn from a lead’s main sheet, and the cover posters, which are drawn from your leads’ sheets. OpenAI says this is the model that holds a reference subject best.
A practical split: draw the cast once on Flare, look at the leads, switch to Sunburst and regenerate only the leads you want to improve. The full comparison is in Sunburst vs Flare.
Writing descriptions that work on GPT Image 2.5
The rules that work on every tier work here. Describe what can be drawn, in plain sentences, in the order face → one anchor feature nobody else has → build and posture → one outfit. Describe causes, not results: a burn scar across the back of the left hand, not a hint of an old injury. Leave lighting and mood out of a sheet — sheets are plain on purpose, because whatever light is on the sheet is carried into every shot.
Write what is there rather than what is not: “no hat” is a weaker instruction than describing the hair. See say what is there, not what is not and describe a face so it is not a generic face.
What it costs
This tier is not covered by the free sheets. Every account’s 15 free sheets are drawn on the default image model, Qwen-Image 3.0, only. On GPT Image 2.5 Sunburst each sheet costs 25 credits from the first one (about $0.16 at the lowest credit price), and the price is printed on the button before you generate. Switch back to the default and whatever free sheets you have left are still there. A generation that is refused is not charged, and most failures are refunded automatically; a few kinds are held until support checks them. Regenerating a sheet is a new charge; the old image stays until you replace it.
For scale: a cast of six leads drawn once each on Sunburst is six images. Most of what a series costs is video seconds, so a better tier for the faces the audience watches most is usually worth it; the rest can stay on the default.
Common errors
What you may meet on these tiers:
| What you see | Usual cause | What to do |
|---|---|---|
| “Blocked by content safety rules — edit the script or description and retry.” | A word or phrase in the description tripped the site’s check, or the model maker’s own filter refused the picture | Edit the description and generate again; nothing was charged |
| “The generation service hit a temporary problem. You weren’t charged — feel free to try again.” | The model was saturated; the site kept the job queued for a while before giving up | Generate again in a few minutes, or switch tier |
| A sheet that ignores part of the description | Too many details competing in one description | Put identity details first; trim decoration |
| Generate All stops partway | Balance ran out; the remaining sheets were not charged | Top up, then generate the rest |
| Credits held for review | A rare failure where the model may already have been paid for the attempt | Nothing to do; support checks it and releases or refunds — see credits held for review |
All the image tiers, side by side
Ten image tiers can be picked in Settings today. The price is per image; the free sheets are on the default only.
| Image tier | Credits per image | In short |
|---|---|---|
| Qwen-Image 3.0 (default) | 7 | The free sheets; a whole cast cheaply |
| Qwen-Image 3.0 Pro (Recommended) | 18 | Leads in close-up; Chinese descriptions |
| SeedDream 5.0 Lite | 8 | 2K locations and props on a budget |
| SeedDream 5.0 Pro | 25 | 2K, photographic skin, long descriptions read to the end |
| Nano Banana 2.1 | 19 | About 2K; Google’s faster model, the cheapest of the six new tiers |
| Nano Banana Pro | 46 | About 2K; detail kept from long descriptions; the most expensive tier |
| GPT Image 2.5 Flare | 25 | About 2K; OpenAI’s faster general-purpose model |
| GPT Image 2.5 Sunburst | 25 | About 2K; OpenAI’s slower, more precise model |
| Grok Imagine Image 2.0 | 35 | About 2K; xAI’s model, strong on lettering |
| MAI-Image-2.6 | 31 | About 1 megapixel; Microsoft’s model, strict content filter |
Which image tier for your sheets turns this into a choice by task.
Sources
- The Rundown: OpenAI ChatGPT Images 2.5 — launch coverage, September 8, 2026; checked October 25, 2026.
- fal: GPT Image 2.5 — the two API models, Flare and Sunburst, as described at launch; checked October 25, 2026.
GPT Image and ChatGPT are trademarks of OpenAI; they are named here to identify the models offered as tiers. Prices on this page are SceneMixer’s own and are read from the live price list.
| GPT Image 2.5 Flare | GPT Image 2.5 Sunburst | |
|---|---|---|
| Shown in Settings as | GPT Image 2.5 Flare | GPT Image 2.5 Sunburst |
| Credits per image | 25 | 25 |
| Output size | About 2K (2,048 px long side) | About 2K (2,048 px long side) |
| OpenAI’s positioning | General purpose, faster | Slower, more precise; holds references |
| Use it for | Supporting cast, locations, props, drafts | Leads, state images, cover posters |
| Free sheets | No — default model only | No — default model only |
FAQ
Can I use GPT Image 2.5 on SceneMixer?
Yes. Both GPT Image 2.5 models are image tiers. Choose GPT Image 2.5 Sunburst or GPT Image 2.5 Flare in Settings, Model Preferences, Image Model, and every new image is drawn on it.
How much does GPT Image 2.5 cost on SceneMixer?
25 credits per image for either model, about $0.16 at the lowest credit price. The price is shown on the button before you generate.
Is GPT Image 2.5 free here?
No. Every account's 15 free sheets are drawn on the default image model, Qwen-Image 3.0. On GPT Image 2.5 each image is charged from the first one.
What is the difference between Sunburst and Flare?
OpenAI built Flare as the faster general-purpose model and Sunburst as the slower, more precise one. On SceneMixer they cost the same and draw at the same size.
Do I need an OpenAI API key or a ChatGPT plan?
No. The tiers are part of SceneMixer and billed in SceneMixer credits.
What resolution does GPT Image 2.5 draw at here?
About 2K: 2,048 pixels on the long side, for example 1152 by 2048 for a vertical image.
Can GPT Image 2.5 make the video?
No. It draws the sheets and cover posters. The shots are made on the video tiers, which use those sheets as references.
Does it keep my characters consistent?
Consistency comes from drawing each character once and reusing that sheet in every shot, on any tier. A tier decides how well the sheet itself matches your description.
Can it paint the title on my cover?
Yes. The promo cover is drawn on the image tier you have selected, and the model paints the title into the poster. Check the spelling before you pick one.
Draw your leads on GPT Image 2.5
Paste a script, pick Sunburst or Flare in Settings, and see the sheets your series will be matched to.
Updated October 25, 2026