GPT Image 1 sits in one workspace with GPT Image 2, Nano Banana Pro, Imagen 4 Ultra, and Seedream—the easiest place to migration-test prompts you already know.

OpenAI · Image Generation API · April 23, 2025

GPT Image 1 — the generation that brought ChatGPT image output to the API

Q: GPT Image 1 or GPT Image 2?

GPT Image 2 is the newer generation and OpenAI's recommended default for new projects: native 4K, flexible sizes, Instant and Thinking dual modes, stronger edit fidelity. GPT Image 1 remains the April 2025 ChatGPT-aligned capability—solid text, quality tiers, mask editing—keep it for compatibility and A/B testing prompts before migration.

Q: Which sizes are supported?

OpenAI API docs primarily list 1024×1024, 1024×1536, and 1536×1024, plus auto. Quality tiers affect fidelity and cost, not pixel dimensions.

Q: What do low / medium / high actually change?

Rendering fidelity, cost, and latency—not canvas size. Higher tiers yield cleaner detail, especially small type. Efficient workflow: explore at low in volume, promote the two or three survivors to high.

Q: Does it support image editing and masks?

Yes. Official image guides document editing with optional alpha-channel masks. One rule to remember: with a mask, describe the full finished image—not just the replaced region.

Q: Didn't this API used to require org verification?

Historically some orgs needed verification before calling gpt-image-1. On iMini you use shared product access—you don't cross that vendor gate yourself.

Q: Can I use outputs commercially?

Follow iMini terms of service and OpenAI usage policy. Confirm rights before paid ads, packaging, or client delivery.

Clear readable in-image text
Photoreal to ink wash in one model
Alpha mask editing supported
Same as ChatGPT, no API key

What is GPT Image 1?

OpenAI's April 2025 image model—the same capability behind ChatGPT image generation, opened to developers as gpt-image-1. Text in posters and UI actually works; alpha mask editing, transparent PNG, and low / medium / high quality tiers you pick per generation. Three sizes, no thinking path: a mature baseline, and the version worth benchmarking before migrating to GPT Image 2.

Developer

OpenAI

API release

April 23, 2025

Output sizes

1024² · 1024×1536 · 1536×1024

Quality tiers

low · medium · high · auto

Edit capability

Image edit + alpha mask

On iMini

Browser · free credits

In-image text you can actually ship

Posters, icon sets, UI mockups, and packaging—letterforms must survive generation. Blurry type voids the whole ad mockup. At small sizes, pull quality to high.

One model, many styles

Photoreal product stills, ink-wash temples, 3D city panoramas, mechanical still life, flat graphic posters—same interface, same prompt habits, switch by changing words.

Scenes that hold up under scrutiny

Solid world knowledge keeps environments, props, and multi-object compositions coherent—hangars look designed, kiosks look installed.

Controls built for production

Size, quality tier, output format and compression, transparency, moderation level, multi-image input, and mask-based editing—the knobs you actually need on a deadline.

What the model actually delivers

Sci-fi hangar, mechanical bouquet, lavender field, isometric city, ink-wash temple, portrait, clean icon poster, airport kiosk UI—each from OpenAI's official documentation gallery.

Compare with today's top models

This 2025 ChatGPT-aligned baseline vs. OpenAI's GPT Image 2, Google's Nano Banana Pro, and Imagen 4 Ultra—where it still holds, and where newer models have moved on.

Dimension	GPT Image 1	GPT Image 2	Nano Banana Pro	Imagen 4 Ultra
Max resolution	Fixed presets, ~1K–1.5K class	Native 4K (3840×2160)	Up to 4K	Up to 2K
In-image text	Strong · posters, UI, packaging	Short copy readable · localization	Multilingual long copy readable	Short copy usable
Edit & mask	Edit + alpha mask	High-fidelity edit + identity hold	Local edit · lighting · DOF · color grade	Text-to-image only
Size flexibility	Three official presets	Any size ≤3:1 (side ×16)	Social to print ratios covered	Standard aspect ratios
Reasoning / Thinking path	No thinking path	Images 2.0 Thinking	Gemini reasoning + web search	No thinking path
Cost / quality control	low / medium / high / auto	low / medium / high	No public quality tiers	Pick by tier: Ultra / standard / Fast
Speed profile	Very fast at quality:low	Instant fast gen + quality tiers	Quality-first Pro channel	Slowest · quality-first
Default pick	Compatibility & legacy alignment	OpenAI default for new projects	Premium typography & final polish	Strictest prompt adherence

Generate free on iMini

Where GPT Image 1 shines

The April 2025 pitch still holds: text, style range, mask editing, transparent backgrounds, and quality tiers you set per generation.

In-image text

Headlines, labels, and UI fields stay readable. At small sizes, quality:high is the line between draft and deliverable—small type collapse makes kiosk and poster concepts unusable.

One interface, many styles

Photoreal, illustration, isometric 3D, ink wash, flat graphics—change style by changing words, not models.

Editing with alpha masks

Mask the region to replace, then describe the full finished image—not just the masked patch. That habit alone fixes most edit failures.

Transparent backgrounds

PNG and WEBP with real alpha channels—product cutouts and sticker sets go straight into compositing pipelines.

Quality tiers, not resolution tiers

low / medium / high / auto trade compute for fidelity without changing canvas size. Explore at low, promote winners to high.

Believable real-world scenes

Hangars, kiosks, storefronts, and brand contexts feel plausible—enough for ad drafts and presentation decks.

One model, eight production workflows

OpenAI's official documentation gallery was built to show exactly these jobs—product, typography, UI, people, and everything in between.

Product & still life

Product shots, mechanical still life, and material-forward hero visuals for e-commerce drafts and packaging concepts.

Posters & icon systems

Ad graphics and icon sets—headline hierarchy and clean shapes must survive rendering. Blurry type voids the poster.

UI & kiosk concepts

Flight information screens, interface mockups, and wayfinding displays with high-contrast panels and readable gate text. Blurry fields make concepts unreviewable.

Portraits & people

Cliffside portraits and editorial people shots for mood boards, decks, and concept frames.

Concept & environment boards

Sci-fi hangars and built environments with enough structural logic for early production reference.

Landscape & nature

Sunrise lavender fields and similar background plates—what used to be a stock photo license fee.

3D & isometric worlds

Miniature city panoramas and isometric scenes for explainers, game pitches, and product metaphors.

Illustration & ink wash

Traditional and painterly textures—misty temples, brush grain—styles photoreal models often flatten.

Generate free on iMini

Teams still on this mature baseline

You wired ChatGPT-aligned generation in 2025, or you want to map gpt-image-1 controls before migrating.

Developers

Official gpt-image-1 surface—size, quality, mask, transparency—without opening an OpenAI org first.

Marketing & growth

Ad stills and campaign concepts in the browser, picking the tier that matches finish level needed.

E-commerce & DTC

Product and packaging visuals, including transparent cutouts ready for compositing.

Designers

Cross-style rapid tests—winners hand off to GPT Image 2 for final resolution and edit fidelity.

Creators

Covers, sticker sets, and series illustration with tiers and cost suited to high update frequency.

Agencies

Mask edits on signed-off frames instead of re-generating whole concepts for every note.

Three steps to run GPT Image 1

Step 1

Open the GPT Image 1 entry

Enter iMini's image workspace and select GPT Image 1—same account and credits as GPT Image 2 and other models.

Step 2

Write what you need clearly

State subject, style, in-image text in quotes, and orientation. For edits describe the full finished image—not just the masked region.

Step 3

Generate, iterate, deliver

Explore at low quality in volume, promote survivors to high, then download—or run the same prompt on GPT Image 2 to see the upgrade gap.

4 ready-to-use prompts

Close to official gallery deliverables. Hover for full text, one-click copy, run directly on iMini. Chinese prompts supported.

Product

Studio product photo of a matte ceramic mug on linen, soft window light, shallow depth of field, brand wordmark on packaging sharp and crisp, commercial catalog style, no watermark.

Poster

Minimal event poster, bold sans-serif headline "SUMMER NIGHTS", indigo gradient, clean geometric shapes, high contrast, print-ready layout, letterforms sharp and readable.

Portrait

Candid portrait on a cliffside trail, golden hour light, 85mm feel, natural skin texture, soft background blur, editorial travel magazine composition.

UI concept

Airport flight information kiosk UI mockup, high-contrast panels, clear wayfinding icons, realistic screen glare, gate and flight text sharp and readable, front product view.

Generate free on iMini