GPT Image 2 generates and edits images through one OpenAI-style API. Async tasks, credit pricing, two variants — production-ready in minutes.
Free text · required · ≤ 32000 chars
auto picks ratio from prompt or refs
4k needs widescreen ratio
Up to 16 · uploaded to R2
Try one of these prompts
Credit-based — 1 credit = $0.001 USD. Pay only for completed generations.
Figures are pre-submit estimates. The final charge uses the actual text-input, image-input, and image-output usage returned by the model.
Reference calculator estimate. For batches it multiplies the per-image subtotal; final billing uses the provider's aggregate usage and does not multiply shared inputs again.
One sample at the cheapest tier (1 image).
GPT Image 2 is OpenAI's image generation and editing model, exposed on reAPI as one async endpoint that covers text-to-image, image-to-image with up to 16 reference URLs, and mask-based inpainting. Prompts run up to 32,000 characters and output picks from 14 aspect ratios at 1K, 2K or 4K — 4K is reserved for the six widescreen ratios (16:9, 9:16, 2:1, 1:2, 21:9, 9:21). Billing is per image.
Real-world workflows and production use cases you can build and ship with this model.
Edit existing e-commerce shots without a reshoot. Swap colorways, change backgrounds, or restyle for a new season — GPT Image 2 keeps the original product untouched while the surrounding scene gets reworked. One image URL goes in, an updated render comes out. Run a hundred SKUs through the same prompt template and the catalog stays visually consistent.
Edit a photo
Spin four hero variants from a single prompt with the stable channel of GPT Image 2 (n=4). Pick the winner, ship to social, iterate on the rest. The basic channel handles single-image drafts when budget matters more than choice. Same endpoint either way — switch with one string in the request body.
Pass up to 16 reference images and the model fuses them into a new composition that respects your brand. Add a mask URL for inpainting — replace a sky, repaint a wall, or swap a logo without redrawing the surrounding pixels. GPT Image 2 handles the fine edges and texture transitions automatically.
Send model `gpt-image-2` for the cheap basic channel — single image, flat per-resolution price. Send `gpt-image-2-official` for the stable channel: batches up to four, quality tiers, inpainting masks, transparent backgrounds, output format and compression control. Switch channels by changing one string. No client refactor, no second SDK to learn.
1 credit equals $0.001 USD. The live estimator on this page costs your exact configuration before you submit, so the GPT Image 2 charge in your dashboard matches what you saw at request time. No metering surprises, no minimum commitment, no monthly platform fee — only what your app actually generates.
Submit a task, get an ID instantly, poll until done. Nothing to time out, nothing to retry. The same async pattern slots into serverless functions, queue workers, and edge runtimes — wherever you run, the GPT Image 2 endpoint fits the shape your stack already uses.
Sign up and grab a key from the dashboard. Free signup credits cover a few hundred basic-channel generations before you ever top up — enough to ship the first feature and watch the bill stay at zero.
OpenPOST a JSON body with `model`, `prompt`, and any optional fields to `/api/v1/images/generations`. Pick `gpt-image-2` for cheap drafts or the stable variant for the full feature set. The endpoint returns a task ID immediately so the request never blocks your handler.
OpenGET `/api/v1/tasks/:id` until status flips to `completed`. The GPT Image 2 response holds image URLs on our CDN — stable for seven days. Mirror them to your own storage if you need them longer than that.
Opencurl -X POST https://reapi.ai/api/v1/images/generations \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-image-2",
"prompt": "a cute red panda eating bamboo, photorealistic",
"size": "1:1",
"resolution": "2k"
}'Rejected synchronously with the failing field named. Check enum values, ranges, and URL fields — nothing is charged.
401 means a missing or invalid Bearer key; 402 means the reserve exceeds your balance. Manage keys and credits in the console.
Reference material and outputs pass automated moderation. Rejected tasks fail with a clear error and a full refund.
A task that reaches a failed state is never charged — the reserve refunds automatically. Keep the task id and retry when ready.
Common questions about this model.
Explore more models in the same category.
OpenAI
From $0.017 per image
Midjourney
From $0.044 per request (4 images)
From $0.030 per image
Black Forest Labs
From $0.028 per image
Try it in the playground or grab an API key to integrate now.