GPT Image 2 API
GPT Image 2 is the proven workhorse of OpenAI's image lineup. Roko API serves it at $0.016 per 1K image, $0.024 per 2K, and $0.032 per 4K with text-to-image, editing, and multi-reference fusion across 14 aspect ratios.
- Public model ID
gpt-image-2- Output tiers
1K / 2K / 4K (1K default)- Billing
Per output by resolution- Endpoint
POST /v1/images/tasks- Catalog snapshot
What is GPT Image 2?
GPT Image 2 is OpenAI's proven image workhorse, covering text-to-image, image-to-image, and multi-reference editing. OpenAI used it as the baseline in the GPT Image 2.5 launch: 2.5 Flare matches its quality at lower latency, and 2.5 Sunburst exceeds it. For workflows already validated on GPT Image 2, OpenAI itself suggests keeping it.
Roko currently exposes 1K, 2K, and 4K tiers plus 14 aspect ratios (auto through 21:9) at $0.016 / $0.024 / $0.032 per image — among the cheapest GPT image tiers in the Roko catalog.
Calls run as async tasks: submit via POST /v1/images/tasks to get a task_id, then poll GET /v1/tasks/{task_id}. The quality field currently exposes auto only; usable parameters follow request validation.
Core capabilities
Full resolution range
1K, 2K, and 4K tiers — 4K at $0.032 per image covers high-resolution needs at this price band.
14 aspect ratios
From 9:21 vertical to 21:9 widescreen, plus 1:2 and 2:1 — stories, squares, and banners in one model.
Multi-reference editing
Pass references to enter image-to-image/editing — fusion and local edits across several inputs.
Proven value tier
Production-proven quality at a price below the 2.5 family — fit for budget-sensitive volume.
GPT Image 2 API pricing
Pricing is generated from the current Roko catalog and billed per output at the chosen resolution. Confirm the console estimate before submitting.
| Resolution | Price | Billing unit |
|---|---|---|
| 1k | $0.016 | per output |
| 2k | $0.024 | per output |
| 4k | $0.032 | per output |
Cost examples
Standard cost for social and web assets — the volume sweet spot.
Print-grade tier; failed images are not billed.
Parameters and supported scope
These fields come from the current public generation_config. quality exposes auto only; capabilities outside the form follow request validation.
| Parameter | Values | Notes |
|---|---|---|
prompt | textarea | Describe the image to create or the changes to apply to references. · Required |
image_urls | url / base64 | Reference images may use URL or Base64. State what must stay and what may change. · Optional |
size | auto / 1:1 / 16:9 / 9:16 / 4:3 / 3:4 / 3:2 / 2:3 / 5:4 / 4:5 / 21:9 / 9:21 / 1:2 / 2:1 | Selects an aspect ratio, resolution tier, or explicit pixel dimensions as configured for the model. · Default: auto · Optional |
quality | auto | The quality tier follows the model config (auto / low / medium / standard / hd) and can directly affect price. · Default: auto · Optional |
resolution | 1K / 2K / 4K | Selects an output resolution tier configured for this model; the value may directly affect price. · Default: 1K · Optional |
Call GPT Image 2 with Roko API
Run the async task flow: submit a task to get a task_id, then poll for the result. Never expose an API key in client-side code.
Submit generation
curl https://api.rokoapi.com/v1/images/tasks \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: YOUR_UNIQUE_REQUEST_ID" \
-d '{
"model": "gpt-image-2",
"prompt": "Minimal flat-illustration banner of a reading nook, warm palette, clean vector shapes",
"size": "16:9",
"resolution": "2K",
"quality": "auto"
}'Query recovery task
curl https://api.rokoapi.com/v1/tasks/$TASK_ID \
-H "Authorization: Bearer $API_KEY"POST /v1/images/tasks returns 202 with a task_id; once generation completes, query GET /v1/tasks/{task_id} for the result or failure reason. Idempotency-Key must be unique; replays with the same key return the original task without re-billing.
Best-fit scenarios
Social and content assets
14 aspect ratios cover vertical, square, and widescreen from one prompt.
Cozy autumn reading corner, flat illustration, warm oranges and creams, 9:16 for stories.
Product visuals and e-commerce
Editing keeps product features intact for fast scene shots.
Place the reference skincare bottle on a sunlit bathroom counter, soft shadows, editorial style.
Volume prototyping and A/B tests
$0.016 per 1K image fits large-scale style exploration.
Four variations of a fitness app onboarding illustration, same style guide, different poses.
4K display and print
Step up to 4K when you need resolution — no external upscaler required.
Conference booth key visual at 4K: abstract gradient waves, brand blue and white, generous negative space.
Important limitations
- quality currently exposes auto only; other values follow request validation.
- It is the previous generation: pick 2.5 Flare for lower latency or 2.5 Sunburst for higher quality.
- A submitted task is not a successful generation — trust the task query result.
- Text, hands, and fine details can still be wrong — review final assets. Output rights depend on your inputs and applicable terms.
GPT Image 2 or the 2.5 family?
If quality is already validated, stay on GPT Image 2 ($0.016 per 1K). Pick 2.5 Flare for ~50% lower latency, or 2.5 Sunburst for higher quality. The forms are nearly identical, so switching is cheap.
All models →GPT Image 2 FAQ
What model ID should I use for GPT Image 2?
Use the public Roko API model ID gpt-image-2 exactly as shown in the model field.
How much does the GPT Image 2 API cost?
The current catalog lists $0.016 per 1K output, $0.024 per 2K, and $0.032 per 4K. Prices can change; confirm the console estimate and final settlement.
GPT Image 2 vs GPT Image 2.5 — what's the difference?
Per OpenAI, 2.5 Flare matches GPT Image 2 quality at roughly half the latency, and 2.5 Sunburst delivers higher quality. If GPT Image 2 already meets your quality bar, OpenAI suggests keeping it and comparing latency after switching.
Which aspect ratios are supported?
Fourteen: auto, 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, 21:9, 9:21, 1:2, and 2:1, with auto as default.
Does it support image editing?
Yes. Passing references (image_urls, URL or Base64) switches the task into image-to-image/editing with local edits and multi-image fusion.
Which endpoint does it use?
Submit async tasks via POST /v1/images/tasks to get a task_id, then poll GET /v1/tasks/{task_id} for the result or failure reason.
Is 4K worth the premium?
1K to 4K costs $0.016 more per image. Web and social usually fit 1K/2K; choose 4K for print, large displays, or assets you plan to crop later.
Does Roko's GPT Image 2 carry official watermarking?
OpenAI applies provenance measures such as C2PA metadata and invisible watermarking to its image products; Roko adds no visible watermark, and outputs follow the live response.