GPT Image 2.5 Flare
A reference-driven OpenAI image model for product preservation, text edits, scene changes, and layered commercial briefs.
重點資訊
| 開發商 | OpenAI |
|---|---|
| 官方模型 ID | gpt-image-2.5-flare |
| 發布日期 | 2026-09-08 |
| Banana Pie 上的供應狀態 | 現已推出 |
| 所需點數 | 每張圖像 25 點數 |
| 參考圖像 | 最多 8 張 |
OpenAI positions GPT Image 2.5 Flare as the fast, everyday default of the GPT Image 2.5 pair, with Sunburst reserved for precision editing (OpenAI announcement, 2026-09-08). In Banana Pie's evaluation it is a practical candidate for reference-heavy commercial image work, especially when a brief combines products, people, text, garments, or scene edits. Its 4K ceiling and support for up to 8 references are useful, but the starting cost and observed run times make it less suited to disposable exploration.
優勢
- Reference-heavy workflows
Support for up to 8 references gives complex briefs room for multiple identities, garments, products, and visual sources.
- Broad editing coverage
Published scenarios include object binding, identity and garment transfer, in-image text replacement, relighting, outpainting, and product preservation across a scene change.
- 4K output ceiling
A maximum resolution of 4K supports work intended for detailed review or larger final delivery.
- Commercial brief coverage
The evaluation suite includes advertising composition, multilingual typography, structured UI graphics, and product material and lighting prompts.
限制
- Consider the wait per run
Published scenario times ranged from 70.5s to 101.0s, so it may feel slow for rapid exploratory prompting.
- Not a low-cost sketch pass
Generations start at 25 credits, which can add up when a concept needs repeated iteration.
- Limited consistency evidence
Each published scenario contains an auto-selected only-run result, leaving repeated-run consistency unmeasured.
- No comparative ranking
The supplied evidence does not include direct model comparisons, so it cannot support claims of superiority over sibling models.
適合選擇 GPT Image 2.5 Flare 的情況
- Your edit depends on preserving a product while changing its setting.
- You need to combine several references for identity, garments, objects, or scene direction.
- Your brief includes typography, text replacement, advertising composition, or structured graphics.
- You need a 4K output ceiling and can budget at least 25 credits per attempt.
- You are handling relighting, outpainting, or other edits that must remain coherent across the scene.
適合考慮其他模型的情況
- You need inexpensive, disposable concept exploration below the 25-credit starting point.
- You need turnaround materially shorter than the published 70.5s to 101.0s observations.
- Your workflow requires more than 8 reference images.
- You need repeated-run consistency or direct comparative evidence before committing credits.
依不同任務，以下模型可能更適合： GPT Image 2.5 Sunburst · Nano Banana 2
實測結果
所有模型都使用相同的固定提示詞，每個模型各發布一項結果，並提供提示詞與設定供你查看。
提示詞
A premium product photograph of exactly one transparent rectangular perfume bottle, half filled with amber liquid, standing upright on wet black stone. Light comes from the upper left, creating coherent refraction, a contact shadow, and one subtle reflection. No text, logo, plants, or extra objects.
設定
- Resolution: 1K
- Aspect Ratio: 1:1
提示詞
Design a clean cream-colored vertical event poster. Show exactly four centered text lines and no other text: "MOONLIGHT MARKET", "月光市集", "18 OCT", and "RIVER HALL". Render "月光市集" in red and all other lines in black.
設定
- Resolution: 1K
- Aspect Ratio: 2:3
提示詞
A photorealistic close-up of an adult potter shaping a clay bowl on a spinning wheel. Both hands are fully visible, each with five natural fingers touching the clay. No other people or hands. Soft window light.
設定
- Resolution: 1K
- Aspect Ratio: 3:4
提示詞
On a matte gray table, exactly three red wooden cubes form a row on the left, exactly two blue glass spheres sit on the right, and one yellow ceramic mug stands centered behind them. The mug handle points right. No other objects or text.
設定
- Resolution: 1K
- Aspect Ratio: 4:3
提示詞
Create a vertical social ad for a fictional running shoe named AERO. Keep the top 15 percent empty. Directly below it, place the headline "RUN LIGHT" in the upper-left. Show exactly one silver shoe in the lower-right, an orange trail curving from the bottom-left, and a round badge reading "42 KM". No other shoes or text.
設定
- Resolution: 1K
- Aspect Ratio: 9:16
提示詞
Create a clean horizontal pricing comparison graphic titled "CHOOSE YOUR PLAN". Use exactly three equal columns labeled "STARTER", "PRO", and "TEAM". Under each column, show exactly three aligned rows labeled "PROJECTS", "STORAGE", and "SUPPORT", followed by one blue button labeled "SELECT". White background, dark navy text, no additional columns or text.
設定
- Resolution: 1K
- Aspect Ratio: 16:9
提示詞
Use Reference 1 as the base tabletop scene. Replace the red mug with the exact dark navy-blue mug from Reference 2 in the same position and orientation. Replace the folded white towel with the exact green-and-white striped napkin from Reference 3 in the same folded area. Replace the green pear with exactly one orange from Reference 4 in the same position. Replace the blue notebook with the exact mustard-yellow hardcover notebook from Reference 5 in the same position. Add exactly the two lemons from Reference 6 to the right of the mug. Preserve the wooden table, camera, framing, wood grain, lighting, shadows, and all other spatial relationships from Reference 1. Do not copy the white product backgrounds from References 2–6, duplicate any asset, or add other objects.
設定
- Resolution: 1K
- Aspect Ratio: 4:3
參考圖像
提示詞
Replace only the sign text with exactly "NIGHT OWL". Preserve the original font style, spacing, perspective, sign material, lighting, and everything else.
設定
- Resolution: 1K
- Aspect Ratio: 3:2
提示詞
Use the portrait in Reference 1 for the person and the isolated jacket in Reference 2 for the garment. Dress the person from Reference 1 in the exact jacket shown in Reference 2. Preserve the person's identity, face, expression, skin, hair, hands, pose, body proportions, background, framing, and lighting from Reference 1. Preserve the jacket's material, color, collar, buttons, pockets, and sleeve patch from Reference 2. Do not copy the ghost mannequin or white product background.
設定
- Resolution: 1K
- Aspect Ratio: 3:4
參考圖像
提示詞
Change the lighting to warm golden-hour sunlight entering from the left window. Do not move, add, remove, or redesign any object. Update highlights and shadows coherently.
設定
- Resolution: 1K
- Aspect Ratio: 16:9
參考圖像
提示詞
Expand the canvas to a 16:9 landscape by naturally continuing the beach, ocean, and sky on both sides. Keep the complete original image centered without cropping, stretching, letterboxing, or modifying it.
設定
- Resolution: 1K
- Aspect Ratio: 16:9
參考圖像
提示詞
Place the sneaker on a wet outdoor basketball court at dusk. Preserve the exact sneaker shape, black geometric side mark, materials, stitching, sole geometry, and camera angle. Add physically coherent contact, reflections, and dusk lighting.
設定
- Resolution: 1K
- Aspect Ratio: 1:1
點數與價格
GPT Image 2.5 Flare starts at 25 credits per generation. In practice, treat that as the minimum cost for each attempt and account for additional runs when refining text, composition, identity, or product details.
常見問題
Is GPT Image 2.5 Flare available now?
Yes. GPT Image 2.5 Flare is currently available in Banana Pie.
How many credits does it use?
Generations start at 25 credits. The exact charge may depend on the settings shown before you run the model.
How many reference images can I use?
You can provide up to 8 references, making it suitable for briefs that combine identity, garments, products, or other source material.
What is the maximum output resolution?
The listed maximum resolution is 4K.
What has Banana Pie evaluated?
Published selections cover typography, anatomy and contact, attribute binding, advertising composition, structured graphics, relighting, outpainting, text replacement, and reference-based edits. Each scenario has only an auto-selected run, so the evidence does not establish repeated-run consistency or superiority over another model.
相關比較
- gpt image 25 flare vs nano banana 2
- gpt image 2 vs gpt image 25 flare
- gpt image 25 flare vs gpt image 25 sunburst
實測文章
測試方式與揭露事項
Conclusions come from a fixed prompt suite and its published selections, covering product materials, typography, anatomy, attribute binding, advertising layouts, structured graphics, reference-based edits, relighting, outpainting, and product preservation; because every selection was marked auto:only-run, the results are treated as workflow evidence rather than a consistency or comparison claim.
Banana Pie 在同一個創作台提供此模型及其他模型的付費使用權。我們的評測結論來自與使用者相同生成流程的測試。