API model or route: gpt-image-2.5-flare-2026-09-08
App catalog ID: gpt-image-2-5-flare
Capabilities & boundaries
What it supports. What the app exposes.
Provider features and app integration are different layers. A feature in the upstream documentation is not automatically enabled in your workspace.
Model and hosting capabilities
Generate images from a text brief or edit supplied images.
The app pins a dated snapshot instead of a rolling model name.
Use the app's format and background controls for the desired deliverable; inspect alpha transparency rather than assuming a white background is transparent.
Before you choose
App-wide attachment checks allow at most five files and 20 MB per file; this route forwards no more than four edit references. Do not confuse the upload limit with the model's edit limit.
Inspect lettering, hands, geometry and factual details before publishing; a detailed prompt does not guarantee an accurate result.
Use source images you have permission to upload. Provider safety rules and usage terms still apply; generated assets are not automatically cleared for every commercial use.
Controls in the audited app integration
App: size, low / medium / high / xhigh / max / auto quality, background, output format, compression and moderation.
App: 1–4 generated images; up to four attached images are forwarded to the edit route.
Put it to work
Start with a more useful visual brief.
Original examples from EZ Ai Assist. Choose supported controls separately, then inspect the generated result.
Workflow 01
Social announcement
Original visual brief. Set any supported size or quality controls in the app separately.
Create a bright square illustration announcing a fictional community garden day. Show raised planting beds, gloves and a watering can, with a clear open area across the top. Use green and yellow accents but no generated lettering or identifiable people.
Workflow 02
Packaging revision
Supply an image you have permission to use; inspect what the model changes.
Using the attached box photograph, replace its red background with a muted lavender studio sweep. Keep the box shape, label artwork and shadow direction recognizable. Leave every printed word unchanged and do not add any props.
Workflow 03
Character concept
Original visual brief. Set any supported size or quality controls in the app separately.
Create an original friendly lighthouse-keeper character wearing a mustard raincoat and round boots, holding a small lantern. Use a tactile clay-animation look against a pale blue background. Show the whole figure with clearly separated hands and feet.
Developer reference
Provider & hosting API pricing
Provider or hosting API reference prices in USD, not EZ Ai Assist subscription prices or app charges. Check the app for your current plan and usage allowance.
GPT Image 2.5 Flare provider reference rates in USD
Request / tier
USD rate
Unit
Text input
$5
per 1M tokens
Cached text input
$1.25
per 1M tokens
Image input
$8
per 1M tokens
Cached image input
$2
per 1M tokens
Image output
$30
per 1M tokens
Standard token rates, not a fixed per-image quote. Inputs, output size and quality affect cost. GPT Image 2.5 can use a different token count than GPT Image 2, so matching token rates do not imply matching image costs.
Common questions
GPT Image 2.5 Flare FAQs
Which API route does GPT Image 2.5 Flare use?
OpenAI Images API: gpt-image-2.5-flare-2026-09-08. The audited app ID is gpt-image-2-5-flare.
Is this currently available in EZ Ai Assist?
The owner-supplied audit lists and routes this choice, but it is not proof of successful live generation. Check your workspace for current access and settings.
Which settings can I actually change in the app?
App: size, low / medium / high / xhigh / max / auto quality, background, output format, compression and moderation. App: 1–4 generated images; up to four attached images are forwarded to the edit route.
Are these the prices I pay in the app?
Provider or hosting API reference prices in USD, not EZ Ai Assist subscription prices or app charges. Check the app for your current plan and usage allowance. Standard token rates, not a fixed per-image quote. Inputs, output size and quality affect cost. GPT Image 2.5 can use a different token count than GPT Image 2, so matching token rates do not imply matching image costs.
What should I check before publishing an image?
Review text, anatomy, geometry, product details and any factual claims. Verify permissions for uploaded references, save a durable copy and inspect the final dimensions and file format. Provider safety rules and terms apply.
Does a prompt unlock unsupported features?
No. Asking for 4K, transparency, editing, multiple images or a seed in prose does not enable an API feature the app does not send. Use the exposed controls and confirm the actual output.
Sources & freshness
Check the original references.
Reviewed . Public provider or hosting documentation supports the reference specifications and prices. App behavior is based on the owner-supplied October 8, 2026 code audit, not live generation tests.