GPT Image 1.5 is the current model behind GPT's image generation and the gpt-image-2 API. Three qualities (Low, Medium, High), three sizes (1024×1024, 1024×1536, 1536×1024), per-image pricing from $0.009. Open the studio below, run a free prompt, and ship production-ready output without setting up API keys.
Join 10,000+ creators shipping with GPT Image 1.5
OpenAI's current flagship image model, replacing DALL·E 3
GPT Image 1.5 is the image model OpenAI currently serves as its flagship. The model id is gpt-image-2 (latest snapshot gpt-image-2-2025-12-16). It accepts text and image input, returns an image plus optional text, and supports three output sizes — 1024×1024, 1024×1536, and 1536×1024 — at Low, Medium, and High quality. DALL·E 2 and DALL·E 3 retire on May 12, 2026 and OpenAI recommends migrating to gpt-image-1, gpt-image-1-mini, or gpt-image-2. This page runs gpt-image-2 directly — open the studio, drop in a prompt, ship the render.
gpt-image-2 high-fidelity quality renders sign text, UI labels, and typography that earlier DALL·E models could not. Per-image high-quality pricing is $0.133 (square) or $0.200 (widescreen and portrait).
1024×1024 square, 1024×1536 portrait, and 1536×1024 landscape. No stitching, no outpainting — the model returns native aspect ratios straight from the API.
Low, Medium, High. Low is $0.009 per 1024×1024 image for drafts and iteration; High is $0.133–$0.200 for finals. Pick quality per render to manage spend.
gpt-image-2 accepts text-only prompts, image-only references, or both in one request. Upload a reference image and describe edits in plain language.
Tier 1 accounts get 5 images per minute; Tier 5 gets 250. OpenAI promotes accounts automatically as spend accrues — no manual quota tickets.
Token pricing: $5/M text in, $10/M text out, $8/M image in, $32/M image out. Per-image flat pricing is available at all three quality levels. Source: developers.openai.com/api/docs/models/gpt-image-2.
Ad copy, posters, and social graphics that hold up at final size
Use gpt-image-2 High quality for ads and posters where text legibility matters. Run drafts at Low ($0.009 per 1024×1024 image) until composition is locked, then regenerate the winner at High for delivery.
What you actually pay per image at each quality and size
Low 1024×1024: $0.009. Medium 1024×1024: $0.034. High 1024×1024: $0.133. Widescreen and portrait at High: $0.200. Token billing separately: $8/M image input, $32/M image output. Source: OpenAI gpt-image-2 model page.
Why OpenAI is retiring DALL·E 3 on May 12, 2026
DALL·E 2 and DALL·E 3 shut down May 12, 2026. OpenAI recommends migrating to gpt-image-1, gpt-image-1-mini, or gpt-image-2. gpt-image-2 brings text + image input in one request, higher quality text rendering, and the three-tier quality/size matrix that DALL·E 3 never had.
Photoreal reference imagery without a photoshoot
Upload a product reference, specify lighting and setting, and gpt-image-2 returns a production-ready shot. Use the 1536×1024 landscape size for hero images, 1024×1536 portrait for mobile commerce.
Four steps, studio and API both covered
Scroll to the generator. gpt-image-2 is the default model — no setup, no API keys, no install. Write your prompt in plain English.
Choose 1024×1024, 1024×1536, or 1536×1024. Pick Low, Medium, or High quality. Low is drafts; High is delivery. Credit cost scales with quality.
Generations land in My Creations. Tweak the prompt, upload a reference image, rerun. History is kept per session so you can trace every iteration.
Download the final render, or use the same prompt on OpenAI's gpt-image-2 API directly — pricing above applies at both tiers.
Pricing, rate limits, sizes, and migration from DALL·E
Free credits on signup, no credit card, no API keys. Open the studio, write a prompt, and ship your first gpt-image-2 render in under a minute.